ONNX export failed with custom image size trained with patchcore model #1290

udasinnayan · 2023-08-21T14:13:59Z

udasinnayan
Aug 21, 2023

Describe the bug

I trained patchcore algorithm with custom image size and want to export the model with ONNX. If I use the patchcore algorithm--> backbone: resnet18 --> with default image size [224, 224] everything works fine without any error with onnx model to tensorrt, but if try to transfer the onnx model that trained with custom image size to tensorrt with I got some error. The error is given into the Logs section. Is the error generated because of custom image size? How do I export the model to ONNX with custom image trained model? Thanks in advance.

Dataset

Folder

Model

PatchCore

Steps to reproduce the behavior

[graphShapeAnalyzer.cpp::processCheck::862] Error Code 4: Internal Error ((Unnamed Layer* 46) [ElementWise]_output: tensor volume exceeds (2^31)-1, dimensions are [65536,13107,256]) In node 45 (parseGraph): INVALID_NODE: Invalid Node - /Sub [graphShapeAnalyzer.cpp::processCheck::862] Error Code 4: Internal Error ((Unnamed Layer* 46) [ElementWise]_output: tensor volume exceeds (2^31)-1, dimensions are [65536,13107,256]) Traceback (most recent call last): File "onnx_export.py", line 39, in <module> raise RuntimeError("ONNX parsing failed. Please check the ONNX model.") RuntimeError: ONNX parsing failed. Please check the ONNX model.

OS information

OS information:

OS: [e.g. Ubuntu 20.04]
Python version: [e.g. 3.8.10]
Anomalib version: [e.g. 0.3.6]
PyTorch version: [e.g. 1.9.0]
CUDA/cuDNN version: [e.g. 11.1]
GPU models and configuration: [e.g. 2x GeForce RTX 3090]
Any other relevant information: [e.g. I'm using a custom dataset]

Expected behavior

Default patchcore model with image size [224, 224] works file without any error from model training to onnx export--> tensorrt. How to export the patchcore model that trained with custom image size to ONNX? Do I need to change the feature extraction code section by any chance?

Screenshots

No response

Pip/GitHub

pip

What version/branch did you use?

1.0.0dev

Configuration YAML

dataset:
  name: demolayer
  format: folder

  task: segmentation # detection, segmentation, classification

  path: /demolayermodel/
  extensions: [.png]
  normal_dir: normal # name of the folder containing normal images.
  abnormal_dir: abnormal # name of the folder containing abnormal images.
  normal_test_dir: normal_test_dir
  mask: /demolayer/mask/


  category: defected
  train_batch_size: 1
  eval_batch_size: 1
  num_workers: 8
  image_size: [2048, 8192] # dimensions to which images are resized (mandatory)
  #center_crop: 256 # dimensions to which images are center-cropped after resizing (optional)
  normalization: imagenet # data distribution to which the images will be normalized: [none, imagenet]
  transform_config:
    train: null
    eval: null
  test_split_mode: from_dir # options: [from_dir, synthetic]
  test_split_ratio: 0.2 # fraction of train images held out testing (usage depends on test_split_mode)
  val_split_mode: same_as_test # options: [same_as_test, from_test, synthetic]
  val_split_ratio: 0.5 # fraction of train/test images held out for validation (usage depends on val_split_mode)
  tiling:
    apply: false
    tile_size: 256
    stride: null
    remove_border_count: 0
    use_random_tiling: true
    random_tile_count: 192

model:
  name: demolayer
  backbone: resnet18 #wide_resnet50_2
  pre_trained: true
  layers:
#    - layer1
#    - layer2
    - layer3
  coreset_sampling_ratio: 0.1
  num_neighbors: 9
  #weight_file:  weights/model.ckpt
  normalization_method: min_max # options: [null, min_max, cdf]

metrics:
  image:
    - F1Score
    - AUROC
  pixel:
    - F1Score
    - AUROC
  threshold:
    method: adaptive #options: [adaptive, manual]
    manual_image: null
    manual_pixel:  null

visualization:
  show_images: False # show images on the screen
  save_images: True # save images to the file system
  log_images: True # log images to the available loggers (if any)
  image_save_path: null # path to which images will be saved
  mode: full # options: ["full", "simple"]

project:
  seed: 0
  path: /results/

logging:
  logger: ["tensorboard"] # options: [comet, tensorboard, wandb, csv] or combinations.
  log_graph: true #true # Logs the model graph to respective logger.

optimization:
  export_mode: onnx # options: onnx, openvino

# PL Trainer Args. Don't add extra parameter here.
trainer:
  enable_checkpointing: true
  default_root_dir: null
  gradient_clip_val: 0
  gradient_clip_algorithm: norm
  num_nodes: 1
  devices: 1 #4
  enable_progress_bar: true
  overfit_batches: 0.0
  track_grad_norm: -1
  check_val_every_n_epoch: 1 # Don't validate before extracting features.
  fast_dev_run: false
  accumulate_grad_batches: 1
  max_epochs: 1
  min_epochs: null
  max_steps: -1
  min_steps: null
  max_time: null
  limit_train_batches: 1.0
  limit_val_batches: 1.0
  limit_test_batches: 1.0
  limit_predict_batches: 1.0
  val_check_interval: 1.0 # Don't validate before extracting features.
  log_every_n_steps: 1
  accelerator: gpu # <"cpu", "gpu", "tpu", "ipu", "hpu", "auto">
  strategy: null # ddp, ddp_spawn
  sync_batchnorm: false
  precision: 32 
  enable_model_summary: true
  num_sanity_val_steps: 0
  profiler: null
  benchmark: false
  deterministic: false
  reload_dataloaders_every_n_epochs: 0
  auto_lr_find: false
  replace_sampler_ddp: true
  detect_anomaly: false
  auto_scale_batch_size: false
  plugins: null
  move_metrics_to_cpu: false
  multiple_trainloader_mode: max_size_cycle

Logs

[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::BatchedNMSDynamic_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::BatchedNMS_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::BatchTilePlugin_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::Clip_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::CoordConvAC version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::CropAndResizeDynamic version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::CropAndResize version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::DecodeBbox3DPlugin version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::DetectionLayer_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::EfficientNMS_Explicit_TF_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::EfficientNMS_Implicit_TF_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::EfficientNMS_ONNX_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::EfficientNMS_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::FlattenConcat_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::GenerateDetection_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::GridAnchor_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::GridAnchorRect_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::InstanceNormalization_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::InstanceNormalization_TRT version 2
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::LReLU_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::ModulatedDeformConv2d version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::MultilevelCropAndResize_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::MultilevelProposeROI_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::MultiscaleDeformableAttnPlugin_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::NMSDynamic_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::NMS_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::Normalize_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::PillarScatterPlugin version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::PriorBox_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::ProposalDynamic version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::ProposalLayer_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::Proposal version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::PyramidROIAlign_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::Region_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::Reorg_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::ResizeNearest_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::ROIAlign_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::RPROI_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::ScatterND version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::SpecialSlice_TRT version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::Split version 1
[08/21/2023-13:24:04] [TRT] [V] Registered plugin creator - ::VoxelGeneratorPlugin version 1
[08/21/2023-13:24:04] [TRT] [V] Adding network input: input with dtype: float32, dimensions: (1, 3, 2048, 8192)
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: input for ONNX tensor: input
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: memory_bank
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: anomaly_map_generator.blur.kernel
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_322
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_323
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_325
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_326
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_328
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_329
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_331
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_332
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_334
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_335
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_337
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_338
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_340
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_341
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_343
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_344
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_346
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_347
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_349
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_350
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_352
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_353
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_355
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_356
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_358
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_359
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_361
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_362
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_364
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Conv_365
[08/21/2023-13:24:04] [TRT] [V] Importing initializer: onnx::Sub_369
[08/21/2023-13:24:04] [TRT] [V] Parsing node: Identity_0 [Identity]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Sub_369
[08/21/2023-13:24:04] [TRT] [V] Identity_0 [Identity] inputs: [onnx::Sub_369 -> (1, 13107, 256)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Registering layer: onnx::Sub_369 for ONNX node: onnx::Sub_369
[08/21/2023-13:24:04] [TRT] [V] Registering layer: Identity_0 for ONNX node: Identity_0
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: onnx::Sub_384 for ONNX tensor: onnx::Sub_384
[08/21/2023-13:24:04] [TRT] [V] Identity_0 [Identity] outputs: [onnx::Sub_384 -> (1, 13107, 256)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/conv1/Conv [Conv]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: input
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_322
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_323
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/conv1/Conv [Conv] inputs: [input -> (1, 3, 2048, 8192)[FLOAT]], [onnx::Conv_322 -> (64, 3, 7, 7)[FLOAT]], [onnx::Conv_323 -> (64)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Convolution input dimensions: (1, 3, 2048, 8192)
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/conv1/Conv for ONNX node: /feature_extractor/feature_extractor/conv1/Conv
[08/21/2023-13:24:04] [TRT] [V] Using kernel: (7, 7), strides: (2, 2), prepadding: (3, 3), postpadding: (3, 3), dilations: (1, 1), numOutputs: 64
[08/21/2023-13:24:04] [TRT] [V] Convolution output dimensions: (1, 64, 1024, 4096)
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/conv1/Conv_output_0 for ONNX tensor: /feature_extractor/feature_extractor/conv1/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/conv1/Conv [Conv] outputs: [/feature_extractor/feature_extractor/conv1/Conv_output_0 -> (1, 64, 1024, 4096)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/act1/Relu [Relu]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/conv1/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/act1/Relu [Relu] inputs: [/feature_extractor/feature_extractor/conv1/Conv_output_0 -> (1, 64, 1024, 4096)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/act1/Relu for ONNX node: /feature_extractor/feature_extractor/act1/Relu
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/act1/Relu_output_0 for ONNX tensor: /feature_extractor/feature_extractor/act1/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/act1/Relu [Relu] outputs: [/feature_extractor/feature_extractor/act1/Relu_output_0 -> (1, 64, 1024, 4096)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/maxpool/MaxPool [MaxPool]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/act1/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/maxpool/MaxPool [MaxPool] inputs: [/feature_extractor/feature_extractor/act1/Relu_output_0 -> (1, 64, 1024, 4096)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/maxpool/MaxPool for ONNX node: /feature_extractor/feature_extractor/maxpool/MaxPool
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/maxpool/MaxPool_output_0 for ONNX tensor: /feature_extractor/feature_extractor/maxpool/MaxPool_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/maxpool/MaxPool [MaxPool] outputs: [/feature_extractor/feature_extractor/maxpool/MaxPool_output_0 -> (1, 64, 512, 2048)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer1/layer1.0/conv1/Conv [Conv]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/maxpool/MaxPool_output_0
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_325
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_326
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer1/layer1.0/conv1/Conv [Conv] inputs: [/feature_extractor/feature_extractor/maxpool/MaxPool_output_0 -> (1, 64, 512, 2048)[FLOAT]], [onnx::Conv_325 -> (64, 64, 3, 3)[FLOAT]], [onnx::Conv_326 -> (64)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Convolution input dimensions: (1, 64, 512, 2048)
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer1/layer1.0/conv1/Conv for ONNX node: /feature_extractor/feature_extractor/layer1/layer1.0/conv1/Conv
[08/21/2023-13:24:04] [TRT] [V] Using kernel: (3, 3), strides: (1, 1), prepadding: (1, 1), postpadding: (1, 1), dilations: (1, 1), numOutputs: 64
[08/21/2023-13:24:04] [TRT] [V] Convolution output dimensions: (1, 64, 512, 2048)
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer1/layer1.0/conv1/Conv_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer1/layer1.0/conv1/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer1/layer1.0/conv1/Conv [Conv] outputs: [/feature_extractor/feature_extractor/layer1/layer1.0/conv1/Conv_output_0 -> (1, 64, 512, 2048)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer1/layer1.0/act1/Relu [Relu]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer1/layer1.0/conv1/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer1/layer1.0/act1/Relu [Relu] inputs: [/feature_extractor/feature_extractor/layer1/layer1.0/conv1/Conv_output_0 -> (1, 64, 512, 2048)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer1/layer1.0/act1/Relu for ONNX node: /feature_extractor/feature_extractor/layer1/layer1.0/act1/Relu
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer1/layer1.0/act1/Relu_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer1/layer1.0/act1/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer1/layer1.0/act1/Relu [Relu] outputs: [/feature_extractor/feature_extractor/layer1/layer1.0/act1/Relu_output_0 -> (1, 64, 512, 2048)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer1/layer1.0/conv2/Conv [Conv]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer1/layer1.0/act1/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_328
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_329
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer1/layer1.0/conv2/Conv [Conv] inputs: [/feature_extractor/feature_extractor/layer1/layer1.0/act1/Relu_output_0 -> (1, 64, 512, 2048)[FLOAT]], [onnx::Conv_328 -> (64, 64, 3, 3)[FLOAT]], [onnx::Conv_329 -> (64)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Convolution input dimensions: (1, 64, 512, 2048)
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer1/layer1.0/conv2/Conv for ONNX node: /feature_extractor/feature_extractor/layer1/layer1.0/conv2/Conv
[08/21/2023-13:24:04] [TRT] [V] Using kernel: (3, 3), strides: (1, 1), prepadding: (1, 1), postpadding: (1, 1), dilations: (1, 1), numOutputs: 64
[08/21/2023-13:24:04] [TRT] [V] Convolution output dimensions: (1, 64, 512, 2048)
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer1/layer1.0/conv2/Conv_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer1/layer1.0/conv2/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer1/layer1.0/conv2/Conv [Conv] outputs: [/feature_extractor/feature_extractor/layer1/layer1.0/conv2/Conv_output_0 -> (1, 64, 512, 2048)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer1/layer1.0/Add [Add]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer1/layer1.0/conv2/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/maxpool/MaxPool_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer1/layer1.0/Add [Add] inputs: [/feature_extractor/feature_extractor/layer1/layer1.0/conv2/Conv_output_0 -> (1, 64, 512, 2048)[FLOAT]], [/feature_extractor/feature_extractor/maxpool/MaxPool_output_0 -> (1, 64, 512, 2048)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer1/layer1.0/Add for ONNX node: /feature_extractor/feature_extractor/layer1/layer1.0/Add
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer1/layer1.0/Add_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer1/layer1.0/Add_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer1/layer1.0/Add [Add] outputs: [/feature_extractor/feature_extractor/layer1/layer1.0/Add_output_0 -> (1, 64, 512, 2048)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer1/layer1.0/act2/Relu [Relu]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer1/layer1.0/Add_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer1/layer1.0/act2/Relu [Relu] inputs: [/feature_extractor/feature_extractor/layer1/layer1.0/Add_output_0 -> (1, 64, 512, 2048)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer1/layer1.0/act2/Relu for ONNX node: /feature_extractor/feature_extractor/layer1/layer1.0/act2/Relu
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer1/layer1.0/act2/Relu_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer1/layer1.0/act2/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer1/layer1.0/act2/Relu [Relu] outputs: [/feature_extractor/feature_extractor/layer1/layer1.0/act2/Relu_output_0 -> (1, 64, 512, 2048)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer1/layer1.1/conv1/Conv [Conv]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer1/layer1.0/act2/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_331
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_332
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer1/layer1.1/conv1/Conv [Conv] inputs: [/feature_extractor/feature_extractor/layer1/layer1.0/act2/Relu_output_0 -> (1, 64, 512, 2048)[FLOAT]], [onnx::Conv_331 -> (64, 64, 3, 3)[FLOAT]], [onnx::Conv_332 -> (64)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Convolution input dimensions: (1, 64, 512, 2048)
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer1/layer1.1/conv1/Conv for ONNX node: /feature_extractor/feature_extractor/layer1/layer1.1/conv1/Conv
[08/21/2023-13:24:04] [TRT] [V] Using kernel: (3, 3), strides: (1, 1), prepadding: (1, 1), postpadding: (1, 1), dilations: (1, 1), numOutputs: 64
[08/21/2023-13:24:04] [TRT] [V] Convolution output dimensions: (1, 64, 512, 2048)
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer1/layer1.1/conv1/Conv_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer1/layer1.1/conv1/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer1/layer1.1/conv1/Conv [Conv] outputs: [/feature_extractor/feature_extractor/layer1/layer1.1/conv1/Conv_output_0 -> (1, 64, 512, 2048)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer1/layer1.1/act1/Relu [Relu]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer1/layer1.1/conv1/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer1/layer1.1/act1/Relu [Relu] inputs: [/feature_extractor/feature_extractor/layer1/layer1.1/conv1/Conv_output_0 -> (1, 64, 512, 2048)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer1/layer1.1/act1/Relu for ONNX node: /feature_extractor/feature_extractor/layer1/layer1.1/act1/Relu
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer1/layer1.1/act1/Relu_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer1/layer1.1/act1/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer1/layer1.1/act1/Relu [Relu] outputs: [/feature_extractor/feature_extractor/layer1/layer1.1/act1/Relu_output_0 -> (1, 64, 512, 2048)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer1/layer1.1/conv2/Conv [Conv]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer1/layer1.1/act1/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_334
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_335
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer1/layer1.1/conv2/Conv [Conv] inputs: [/feature_extractor/feature_extractor/layer1/layer1.1/act1/Relu_output_0 -> (1, 64, 512, 2048)[FLOAT]], [onnx::Conv_334 -> (64, 64, 3, 3)[FLOAT]], [onnx::Conv_335 -> (64)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Convolution input dimensions: (1, 64, 512, 2048)
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer1/layer1.1/conv2/Conv for ONNX node: /feature_extractor/feature_extractor/layer1/layer1.1/conv2/Conv
[08/21/2023-13:24:04] [TRT] [V] Using kernel: (3, 3), strides: (1, 1), prepadding: (1, 1), postpadding: (1, 1), dilations: (1, 1), numOutputs: 64
[08/21/2023-13:24:04] [TRT] [V] Convolution output dimensions: (1, 64, 512, 2048)
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer1/layer1.1/conv2/Conv_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer1/layer1.1/conv2/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer1/layer1.1/conv2/Conv [Conv] outputs: [/feature_extractor/feature_extractor/layer1/layer1.1/conv2/Conv_output_0 -> (1, 64, 512, 2048)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer1/layer1.1/Add [Add]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer1/layer1.1/conv2/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer1/layer1.0/act2/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer1/layer1.1/Add [Add] inputs: [/feature_extractor/feature_extractor/layer1/layer1.1/conv2/Conv_output_0 -> (1, 64, 512, 2048)[FLOAT]], [/feature_extractor/feature_extractor/layer1/layer1.0/act2/Relu_output_0 -> (1, 64, 512, 2048)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer1/layer1.1/Add for ONNX node: /feature_extractor/feature_extractor/layer1/layer1.1/Add
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer1/layer1.1/Add_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer1/layer1.1/Add_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer1/layer1.1/Add [Add] outputs: [/feature_extractor/feature_extractor/layer1/layer1.1/Add_output_0 -> (1, 64, 512, 2048)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer1/layer1.1/act2/Relu [Relu]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer1/layer1.1/Add_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer1/layer1.1/act2/Relu [Relu] inputs: [/feature_extractor/feature_extractor/layer1/layer1.1/Add_output_0 -> (1, 64, 512, 2048)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer1/layer1.1/act2/Relu for ONNX node: /feature_extractor/feature_extractor/layer1/layer1.1/act2/Relu
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer1/layer1.1/act2/Relu_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer1/layer1.1/act2/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer1/layer1.1/act2/Relu [Relu] outputs: [/feature_extractor/feature_extractor/layer1/layer1.1/act2/Relu_output_0 -> (1, 64, 512, 2048)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer2/layer2.0/conv1/Conv [Conv]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer1/layer1.1/act2/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_337
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_338
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer2/layer2.0/conv1/Conv [Conv] inputs: [/feature_extractor/feature_extractor/layer1/layer1.1/act2/Relu_output_0 -> (1, 64, 512, 2048)[FLOAT]], [onnx::Conv_337 -> (128, 64, 3, 3)[FLOAT]], [onnx::Conv_338 -> (128)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Convolution input dimensions: (1, 64, 512, 2048)
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer2/layer2.0/conv1/Conv for ONNX node: /feature_extractor/feature_extractor/layer2/layer2.0/conv1/Conv
[08/21/2023-13:24:04] [TRT] [V] Using kernel: (3, 3), strides: (2, 2), prepadding: (1, 1), postpadding: (1, 1), dilations: (1, 1), numOutputs: 128
[08/21/2023-13:24:04] [TRT] [V] Convolution output dimensions: (1, 128, 256, 1024)
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer2/layer2.0/conv1/Conv_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer2/layer2.0/conv1/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer2/layer2.0/conv1/Conv [Conv] outputs: [/feature_extractor/feature_extractor/layer2/layer2.0/conv1/Conv_output_0 -> (1, 128, 256, 1024)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer2/layer2.0/act1/Relu [Relu]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer2/layer2.0/conv1/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer2/layer2.0/act1/Relu [Relu] inputs: [/feature_extractor/feature_extractor/layer2/layer2.0/conv1/Conv_output_0 -> (1, 128, 256, 1024)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer2/layer2.0/act1/Relu for ONNX node: /feature_extractor/feature_extractor/layer2/layer2.0/act1/Relu
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer2/layer2.0/act1/Relu_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer2/layer2.0/act1/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer2/layer2.0/act1/Relu [Relu] outputs: [/feature_extractor/feature_extractor/layer2/layer2.0/act1/Relu_output_0 -> (1, 128, 256, 1024)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer2/layer2.0/conv2/Conv [Conv]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer2/layer2.0/act1/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_340
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_341
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer2/layer2.0/conv2/Conv [Conv] inputs: [/feature_extractor/feature_extractor/layer2/layer2.0/act1/Relu_output_0 -> (1, 128, 256, 1024)[FLOAT]], [onnx::Conv_340 -> (128, 128, 3, 3)[FLOAT]], [onnx::Conv_341 -> (128)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Convolution input dimensions: (1, 128, 256, 1024)
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer2/layer2.0/conv2/Conv for ONNX node: /feature_extractor/feature_extractor/layer2/layer2.0/conv2/Conv
[08/21/2023-13:24:04] [TRT] [V] Using kernel: (3, 3), strides: (1, 1), prepadding: (1, 1), postpadding: (1, 1), dilations: (1, 1), numOutputs: 128
[08/21/2023-13:24:04] [TRT] [V] Convolution output dimensions: (1, 128, 256, 1024)
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer2/layer2.0/conv2/Conv_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer2/layer2.0/conv2/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer2/layer2.0/conv2/Conv [Conv] outputs: [/feature_extractor/feature_extractor/layer2/layer2.0/conv2/Conv_output_0 -> (1, 128, 256, 1024)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer2/layer2.0/downsample/downsample.0/Conv [Conv]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer1/layer1.1/act2/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_343
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_344
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer2/layer2.0/downsample/downsample.0/Conv [Conv] inputs: [/feature_extractor/feature_extractor/layer1/layer1.1/act2/Relu_output_0 -> (1, 64, 512, 2048)[FLOAT]], [onnx::Conv_343 -> (128, 64, 1, 1)[FLOAT]], [onnx::Conv_344 -> (128)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Convolution input dimensions: (1, 64, 512, 2048)
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer2/layer2.0/downsample/downsample.0/Conv for ONNX node: /feature_extractor/feature_extractor/layer2/layer2.0/downsample/downsample.0/Conv
[08/21/2023-13:24:04] [TRT] [V] Using kernel: (1, 1), strides: (2, 2), prepadding: (0, 0), postpadding: (0, 0), dilations: (1, 1), numOutputs: 128
[08/21/2023-13:24:04] [TRT] [V] Convolution output dimensions: (1, 128, 256, 1024)
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer2/layer2.0/downsample/downsample.0/Conv_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer2/layer2.0/downsample/downsample.0/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer2/layer2.0/downsample/downsample.0/Conv [Conv] outputs: [/feature_extractor/feature_extractor/layer2/layer2.0/downsample/downsample.0/Conv_output_0 -> (1, 128, 256, 1024)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer2/layer2.0/Add [Add]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer2/layer2.0/conv2/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer2/layer2.0/downsample/downsample.0/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer2/layer2.0/Add [Add] inputs: [/feature_extractor/feature_extractor/layer2/layer2.0/conv2/Conv_output_0 -> (1, 128, 256, 1024)[FLOAT]], [/feature_extractor/feature_extractor/layer2/layer2.0/downsample/downsample.0/Conv_output_0 -> (1, 128, 256, 1024)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer2/layer2.0/Add for ONNX node: /feature_extractor/feature_extractor/layer2/layer2.0/Add
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer2/layer2.0/Add_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer2/layer2.0/Add_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer2/layer2.0/Add [Add] outputs: [/feature_extractor/feature_extractor/layer2/layer2.0/Add_output_0 -> (1, 128, 256, 1024)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer2/layer2.0/act2/Relu [Relu]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer2/layer2.0/Add_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer2/layer2.0/act2/Relu [Relu] inputs: [/feature_extractor/feature_extractor/layer2/layer2.0/Add_output_0 -> (1, 128, 256, 1024)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer2/layer2.0/act2/Relu for ONNX node: /feature_extractor/feature_extractor/layer2/layer2.0/act2/Relu
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer2/layer2.0/act2/Relu_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer2/layer2.0/act2/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer2/layer2.0/act2/Relu [Relu] outputs: [/feature_extractor/feature_extractor/layer2/layer2.0/act2/Relu_output_0 -> (1, 128, 256, 1024)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer2/layer2.1/conv1/Conv [Conv]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer2/layer2.0/act2/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_346
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_347
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer2/layer2.1/conv1/Conv [Conv] inputs: [/feature_extractor/feature_extractor/layer2/layer2.0/act2/Relu_output_0 -> (1, 128, 256, 1024)[FLOAT]], [onnx::Conv_346 -> (128, 128, 3, 3)[FLOAT]], [onnx::Conv_347 -> (128)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Convolution input dimensions: (1, 128, 256, 1024)
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer2/layer2.1/conv1/Conv for ONNX node: /feature_extractor/feature_extractor/layer2/layer2.1/conv1/Conv
[08/21/2023-13:24:04] [TRT] [V] Using kernel: (3, 3), strides: (1, 1), prepadding: (1, 1), postpadding: (1, 1), dilations: (1, 1), numOutputs: 128
[08/21/2023-13:24:04] [TRT] [V] Convolution output dimensions: (1, 128, 256, 1024)
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer2/layer2.1/conv1/Conv_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer2/layer2.1/conv1/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer2/layer2.1/conv1/Conv [Conv] outputs: [/feature_extractor/feature_extractor/layer2/layer2.1/conv1/Conv_output_0 -> (1, 128, 256, 1024)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer2/layer2.1/act1/Relu [Relu]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer2/layer2.1/conv1/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer2/layer2.1/act1/Relu [Relu] inputs: [/feature_extractor/feature_extractor/layer2/layer2.1/conv1/Conv_output_0 -> (1, 128, 256, 1024)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer2/layer2.1/act1/Relu for ONNX node: /feature_extractor/feature_extractor/layer2/layer2.1/act1/Relu
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer2/layer2.1/act1/Relu_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer2/layer2.1/act1/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer2/layer2.1/act1/Relu [Relu] outputs: [/feature_extractor/feature_extractor/layer2/layer2.1/act1/Relu_output_0 -> (1, 128, 256, 1024)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer2/layer2.1/conv2/Conv [Conv]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer2/layer2.1/act1/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_349
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_350
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer2/layer2.1/conv2/Conv [Conv] inputs: [/feature_extractor/feature_extractor/layer2/layer2.1/act1/Relu_output_0 -> (1, 128, 256, 1024)[FLOAT]], [onnx::Conv_349 -> (128, 128, 3, 3)[FLOAT]], [onnx::Conv_350 -> (128)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Convolution input dimensions: (1, 128, 256, 1024)
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer2/layer2.1/conv2/Conv for ONNX node: /feature_extractor/feature_extractor/layer2/layer2.1/conv2/Conv
[08/21/2023-13:24:04] [TRT] [V] Using kernel: (3, 3), strides: (1, 1), prepadding: (1, 1), postpadding: (1, 1), dilations: (1, 1), numOutputs: 128
[08/21/2023-13:24:04] [TRT] [V] Convolution output dimensions: (1, 128, 256, 1024)
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer2/layer2.1/conv2/Conv_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer2/layer2.1/conv2/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer2/layer2.1/conv2/Conv [Conv] outputs: [/feature_extractor/feature_extractor/layer2/layer2.1/conv2/Conv_output_0 -> (1, 128, 256, 1024)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer2/layer2.1/Add [Add]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer2/layer2.1/conv2/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer2/layer2.0/act2/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer2/layer2.1/Add [Add] inputs: [/feature_extractor/feature_extractor/layer2/layer2.1/conv2/Conv_output_0 -> (1, 128, 256, 1024)[FLOAT]], [/feature_extractor/feature_extractor/layer2/layer2.0/act2/Relu_output_0 -> (1, 128, 256, 1024)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer2/layer2.1/Add for ONNX node: /feature_extractor/feature_extractor/layer2/layer2.1/Add
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer2/layer2.1/Add_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer2/layer2.1/Add_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer2/layer2.1/Add [Add] outputs: [/feature_extractor/feature_extractor/layer2/layer2.1/Add_output_0 -> (1, 128, 256, 1024)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer2/layer2.1/act2/Relu [Relu]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer2/layer2.1/Add_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer2/layer2.1/act2/Relu [Relu] inputs: [/feature_extractor/feature_extractor/layer2/layer2.1/Add_output_0 -> (1, 128, 256, 1024)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer2/layer2.1/act2/Relu for ONNX node: /feature_extractor/feature_extractor/layer2/layer2.1/act2/Relu
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer2/layer2.1/act2/Relu_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer2/layer2.1/act2/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer2/layer2.1/act2/Relu [Relu] outputs: [/feature_extractor/feature_extractor/layer2/layer2.1/act2/Relu_output_0 -> (1, 128, 256, 1024)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer3/layer3.0/conv1/Conv [Conv]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer2/layer2.1/act2/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_352
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_353
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer3/layer3.0/conv1/Conv [Conv] inputs: [/feature_extractor/feature_extractor/layer2/layer2.1/act2/Relu_output_0 -> (1, 128, 256, 1024)[FLOAT]], [onnx::Conv_352 -> (256, 128, 3, 3)[FLOAT]], [onnx::Conv_353 -> (256)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Convolution input dimensions: (1, 128, 256, 1024)
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer3/layer3.0/conv1/Conv for ONNX node: /feature_extractor/feature_extractor/layer3/layer3.0/conv1/Conv
[08/21/2023-13:24:04] [TRT] [V] Using kernel: (3, 3), strides: (2, 2), prepadding: (1, 1), postpadding: (1, 1), dilations: (1, 1), numOutputs: 256
[08/21/2023-13:24:04] [TRT] [V] Convolution output dimensions: (1, 256, 128, 512)
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer3/layer3.0/conv1/Conv_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer3/layer3.0/conv1/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer3/layer3.0/conv1/Conv [Conv] outputs: [/feature_extractor/feature_extractor/layer3/layer3.0/conv1/Conv_output_0 -> (1, 256, 128, 512)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer3/layer3.0/act1/Relu [Relu]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer3/layer3.0/conv1/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer3/layer3.0/act1/Relu [Relu] inputs: [/feature_extractor/feature_extractor/layer3/layer3.0/conv1/Conv_output_0 -> (1, 256, 128, 512)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer3/layer3.0/act1/Relu for ONNX node: /feature_extractor/feature_extractor/layer3/layer3.0/act1/Relu
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer3/layer3.0/act1/Relu_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer3/layer3.0/act1/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer3/layer3.0/act1/Relu [Relu] outputs: [/feature_extractor/feature_extractor/layer3/layer3.0/act1/Relu_output_0 -> (1, 256, 128, 512)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer3/layer3.0/conv2/Conv [Conv]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer3/layer3.0/act1/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_355
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_356
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer3/layer3.0/conv2/Conv [Conv] inputs: [/feature_extractor/feature_extractor/layer3/layer3.0/act1/Relu_output_0 -> (1, 256, 128, 512)[FLOAT]], [onnx::Conv_355 -> (256, 256, 3, 3)[FLOAT]], [onnx::Conv_356 -> (256)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Convolution input dimensions: (1, 256, 128, 512)
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer3/layer3.0/conv2/Conv for ONNX node: /feature_extractor/feature_extractor/layer3/layer3.0/conv2/Conv
[08/21/2023-13:24:04] [TRT] [V] Using kernel: (3, 3), strides: (1, 1), prepadding: (1, 1), postpadding: (1, 1), dilations: (1, 1), numOutputs: 256
[08/21/2023-13:24:04] [TRT] [V] Convolution output dimensions: (1, 256, 128, 512)
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer3/layer3.0/conv2/Conv_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer3/layer3.0/conv2/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer3/layer3.0/conv2/Conv [Conv] outputs: [/feature_extractor/feature_extractor/layer3/layer3.0/conv2/Conv_output_0 -> (1, 256, 128, 512)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer3/layer3.0/downsample/downsample.0/Conv [Conv]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer2/layer2.1/act2/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_358
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_359
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer3/layer3.0/downsample/downsample.0/Conv [Conv] inputs: [/feature_extractor/feature_extractor/layer2/layer2.1/act2/Relu_output_0 -> (1, 128, 256, 1024)[FLOAT]], [onnx::Conv_358 -> (256, 128, 1, 1)[FLOAT]], [onnx::Conv_359 -> (256)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Convolution input dimensions: (1, 128, 256, 1024)
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer3/layer3.0/downsample/downsample.0/Conv for ONNX node: /feature_extractor/feature_extractor/layer3/layer3.0/downsample/downsample.0/Conv
[08/21/2023-13:24:04] [TRT] [V] Using kernel: (1, 1), strides: (2, 2), prepadding: (0, 0), postpadding: (0, 0), dilations: (1, 1), numOutputs: 256
[08/21/2023-13:24:04] [TRT] [V] Convolution output dimensions: (1, 256, 128, 512)
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer3/layer3.0/downsample/downsample.0/Conv_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer3/layer3.0/downsample/downsample.0/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer3/layer3.0/downsample/downsample.0/Conv [Conv] outputs: [/feature_extractor/feature_extractor/layer3/layer3.0/downsample/downsample.0/Conv_output_0 -> (1, 256, 128, 512)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer3/layer3.0/Add [Add]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer3/layer3.0/conv2/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer3/layer3.0/downsample/downsample.0/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer3/layer3.0/Add [Add] inputs: [/feature_extractor/feature_extractor/layer3/layer3.0/conv2/Conv_output_0 -> (1, 256, 128, 512)[FLOAT]], [/feature_extractor/feature_extractor/layer3/layer3.0/downsample/downsample.0/Conv_output_0 -> (1, 256, 128, 512)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer3/layer3.0/Add for ONNX node: /feature_extractor/feature_extractor/layer3/layer3.0/Add
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer3/layer3.0/Add_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer3/layer3.0/Add_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer3/layer3.0/Add [Add] outputs: [/feature_extractor/feature_extractor/layer3/layer3.0/Add_output_0 -> (1, 256, 128, 512)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer3/layer3.0/act2/Relu [Relu]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer3/layer3.0/Add_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer3/layer3.0/act2/Relu [Relu] inputs: [/feature_extractor/feature_extractor/layer3/layer3.0/Add_output_0 -> (1, 256, 128, 512)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer3/layer3.0/act2/Relu for ONNX node: /feature_extractor/feature_extractor/layer3/layer3.0/act2/Relu
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer3/layer3.0/act2/Relu_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer3/layer3.0/act2/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer3/layer3.0/act2/Relu [Relu] outputs: [/feature_extractor/feature_extractor/layer3/layer3.0/act2/Relu_output_0 -> (1, 256, 128, 512)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer3/layer3.1/conv1/Conv [Conv]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer3/layer3.0/act2/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_361
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_362
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer3/layer3.1/conv1/Conv [Conv] inputs: [/feature_extractor/feature_extractor/layer3/layer3.0/act2/Relu_output_0 -> (1, 256, 128, 512)[FLOAT]], [onnx::Conv_361 -> (256, 256, 3, 3)[FLOAT]], [onnx::Conv_362 -> (256)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Convolution input dimensions: (1, 256, 128, 512)
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer3/layer3.1/conv1/Conv for ONNX node: /feature_extractor/feature_extractor/layer3/layer3.1/conv1/Conv
[08/21/2023-13:24:04] [TRT] [V] Using kernel: (3, 3), strides: (1, 1), prepadding: (1, 1), postpadding: (1, 1), dilations: (1, 1), numOutputs: 256
[08/21/2023-13:24:04] [TRT] [V] Convolution output dimensions: (1, 256, 128, 512)
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer3/layer3.1/conv1/Conv_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer3/layer3.1/conv1/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer3/layer3.1/conv1/Conv [Conv] outputs: [/feature_extractor/feature_extractor/layer3/layer3.1/conv1/Conv_output_0 -> (1, 256, 128, 512)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer3/layer3.1/act1/Relu [Relu]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer3/layer3.1/conv1/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer3/layer3.1/act1/Relu [Relu] inputs: [/feature_extractor/feature_extractor/layer3/layer3.1/conv1/Conv_output_0 -> (1, 256, 128, 512)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer3/layer3.1/act1/Relu for ONNX node: /feature_extractor/feature_extractor/layer3/layer3.1/act1/Relu
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer3/layer3.1/act1/Relu_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer3/layer3.1/act1/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer3/layer3.1/act1/Relu [Relu] outputs: [/feature_extractor/feature_extractor/layer3/layer3.1/act1/Relu_output_0 -> (1, 256, 128, 512)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer3/layer3.1/conv2/Conv [Conv]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer3/layer3.1/act1/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_364
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Conv_365
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer3/layer3.1/conv2/Conv [Conv] inputs: [/feature_extractor/feature_extractor/layer3/layer3.1/act1/Relu_output_0 -> (1, 256, 128, 512)[FLOAT]], [onnx::Conv_364 -> (256, 256, 3, 3)[FLOAT]], [onnx::Conv_365 -> (256)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Convolution input dimensions: (1, 256, 128, 512)
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer3/layer3.1/conv2/Conv for ONNX node: /feature_extractor/feature_extractor/layer3/layer3.1/conv2/Conv
[08/21/2023-13:24:04] [TRT] [V] Using kernel: (3, 3), strides: (1, 1), prepadding: (1, 1), postpadding: (1, 1), dilations: (1, 1), numOutputs: 256
[08/21/2023-13:24:04] [TRT] [V] Convolution output dimensions: (1, 256, 128, 512)
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer3/layer3.1/conv2/Conv_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer3/layer3.1/conv2/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer3/layer3.1/conv2/Conv [Conv] outputs: [/feature_extractor/feature_extractor/layer3/layer3.1/conv2/Conv_output_0 -> (1, 256, 128, 512)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer3/layer3.1/Add [Add]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer3/layer3.1/conv2/Conv_output_0
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer3/layer3.0/act2/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer3/layer3.1/Add [Add] inputs: [/feature_extractor/feature_extractor/layer3/layer3.1/conv2/Conv_output_0 -> (1, 256, 128, 512)[FLOAT]], [/feature_extractor/feature_extractor/layer3/layer3.0/act2/Relu_output_0 -> (1, 256, 128, 512)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer3/layer3.1/Add for ONNX node: /feature_extractor/feature_extractor/layer3/layer3.1/Add
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer3/layer3.1/Add_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer3/layer3.1/Add_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer3/layer3.1/Add [Add] outputs: [/feature_extractor/feature_extractor/layer3/layer3.1/Add_output_0 -> (1, 256, 128, 512)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_extractor/feature_extractor/layer3/layer3.1/act2/Relu [Relu]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer3/layer3.1/Add_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer3/layer3.1/act2/Relu [Relu] inputs: [/feature_extractor/feature_extractor/layer3/layer3.1/Add_output_0 -> (1, 256, 128, 512)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_extractor/feature_extractor/layer3/layer3.1/act2/Relu for ONNX node: /feature_extractor/feature_extractor/layer3/layer3.1/act2/Relu
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_extractor/feature_extractor/layer3/layer3.1/act2/Relu_output_0 for ONNX tensor: /feature_extractor/feature_extractor/layer3/layer3.1/act2/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_extractor/feature_extractor/layer3/layer3.1/act2/Relu [Relu] outputs: [/feature_extractor/feature_extractor/layer3/layer3.1/act2/Relu_output_0 -> (1, 256, 128, 512)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_pooler/Constant [Constant]
[08/21/2023-13:24:04] [TRT] [V] /feature_pooler/Constant [Constant] inputs: 
[08/21/2023-13:24:04] [TRT] [W] onnx2trt_utils.cpp:374: Your ONNX model has been generated with INT64 weights, while TensorRT does not natively support INT64. Attempting to cast down to INT32.
[08/21/2023-13:24:04] [TRT] [V] /feature_pooler/Constant [Constant] outputs: [/feature_pooler/Constant_output_0 -> (8)[INT32]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_pooler/Pad [Pad]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_extractor/feature_extractor/layer3/layer3.1/act2/Relu_output_0
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_pooler/Constant_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_pooler/Pad [Pad] inputs: [/feature_extractor/feature_extractor/layer3/layer3.1/act2/Relu_output_0 -> (1, 256, 128, 512)[FLOAT]], [/feature_pooler/Constant_output_0 -> (8)[INT32]], 
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_pooler/Pad for ONNX node: /feature_pooler/Pad
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_pooler/Pad_output_0 for ONNX tensor: /feature_pooler/Pad_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_pooler/Pad [Pad] outputs: [/feature_pooler/Pad_output_0 -> (1, 256, 130, 514)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /feature_pooler/AveragePool [AveragePool]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_pooler/Pad_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_pooler/AveragePool [AveragePool] inputs: [/feature_pooler/Pad_output_0 -> (1, 256, 130, 514)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /feature_pooler/AveragePool for ONNX node: /feature_pooler/AveragePool
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /feature_pooler/AveragePool_output_0 for ONNX tensor: /feature_pooler/AveragePool_output_0
[08/21/2023-13:24:04] [TRT] [V] /feature_pooler/AveragePool [AveragePool] outputs: [/feature_pooler/AveragePool_output_0 -> (1, 256, 128, 512)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /Constant [Constant]
[08/21/2023-13:24:04] [TRT] [V] /Constant [Constant] inputs: 
[08/21/2023-13:24:04] [TRT] [V] /Constant [Constant] outputs: [/Constant_output_0 -> ()[INT32]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /Transpose [Transpose]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /feature_pooler/AveragePool_output_0
[08/21/2023-13:24:04] [TRT] [V] /Transpose [Transpose] inputs: [/feature_pooler/AveragePool_output_0 -> (1, 256, 128, 512)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /Transpose for ONNX node: /Transpose
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /Transpose_output_0 for ONNX tensor: /Transpose_output_0
[08/21/2023-13:24:04] [TRT] [V] /Transpose [Transpose] outputs: [/Transpose_output_0 -> (1, 128, 512, 256)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /Constant_1 [Constant]
[08/21/2023-13:24:04] [TRT] [V] /Constant_1 [Constant] inputs: 
[08/21/2023-13:24:04] [TRT] [V] /Constant_1 [Constant] outputs: [/Constant_1_output_0 -> (2)[INT32]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /Reshape [Reshape]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /Transpose_output_0
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /Constant_1_output_0
[08/21/2023-13:24:04] [TRT] [V] /Reshape [Reshape] inputs: [/Transpose_output_0 -> (1, 128, 512, 256)[FLOAT]], [/Constant_1_output_0 -> (2)[INT32]], 
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /Reshape for ONNX node: /Reshape
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /Reshape_output_0 for ONNX tensor: /Reshape_output_0
[08/21/2023-13:24:04] [TRT] [V] /Reshape [Reshape] outputs: [/Reshape_output_0 -> (65536, 256)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /Unsqueeze [Unsqueeze]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /Reshape_output_0
[08/21/2023-13:24:04] [TRT] [V] /Unsqueeze [Unsqueeze] inputs: [/Reshape_output_0 -> (65536, 256)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Original shape: (65536, 256), unsqueezing to: (65536, 1, 256)
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /Unsqueeze for ONNX node: /Unsqueeze
[08/21/2023-13:24:04] [TRT] [V] Registering tensor: /Unsqueeze_output_0 for ONNX tensor: /Unsqueeze_output_0
[08/21/2023-13:24:04] [TRT] [V] /Unsqueeze [Unsqueeze] outputs: [/Unsqueeze_output_0 -> (65536, 1, 256)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /Constant_2 [Constant]
[08/21/2023-13:24:04] [TRT] [V] /Constant_2 [Constant] inputs: 
[08/21/2023-13:24:04] [TRT] [V] /Constant_2 [Constant] outputs: [/Constant_2_output_0 -> (1)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Parsing node: /Sub [Sub]
[08/21/2023-13:24:04] [TRT] [V] Searching for input: /Unsqueeze_output_0
[08/21/2023-13:24:04] [TRT] [V] Searching for input: onnx::Sub_369
[08/21/2023-13:24:04] [TRT] [V] /Sub [Sub] inputs: [/Unsqueeze_output_0 -> (65536, 1, 256)[FLOAT]], [onnx::Sub_369 -> (1, 13107, 256)[FLOAT]], 
[08/21/2023-13:24:04] [TRT] [V] Registering layer: /Sub for ONNX node: /Sub
[08/21/2023-13:24:04] [TRT] [E] [graphShapeAnalyzer.cpp::processCheck::862] Error Code 4: Internal Error ((Unnamed Layer* 46) [ElementWise]_output: tensor volume exceeds (2^31)-1, dimensions are [65536,13107,256])
In node 45 (parseGraph): INVALID_NODE: Invalid Node - /Sub
[graphShapeAnalyzer.cpp::processCheck::862] Error Code 4: Internal Error ((Unnamed Layer* 46) [ElementWise]_output: tensor volume exceeds (2^31)-1, dimensions are [65536,13107,256])
Traceback (most recent call last):
  File "onnx_export.py", line 39, in <module>
    raise RuntimeError("ONNX parsing failed. Please check the ONNX model.")
RuntimeError: ONNX parsing failed. Please check the ONNX model.

Code of Conduct

I agree to follow this project's Code of Conduct

Answered by blaz-r

Aug 22, 2023

The image is downscaled as part of the procedure. If you want your output to be original size, you could upscale the prediction back to 2048x8192.

View full answer

blaz-r · 2023-08-21T21:02:31Z

blaz-r
Aug 21, 2023

I believe this comes from distance calculation that is done as part of model, where the result is entire matrix of distances. From the error it seems like this exceeds the size that is supported (2^31)-1. I am not familiar with TensorRT so I can't help much in that regard. One thing is though: image size of 2048x8192 is quite huge for patchcore, so I'd say that you try something between the default 224x224 and this large one.

0 replies

udasinnayan · 2023-08-22T09:34:23Z

udasinnayan
Aug 22, 2023
Author

@blaz-r thanks for your reply. It's working with default image size 224*224, but failed with custom image size because of dimensionality, [graphShapeAnalyzer.cpp::processCheck::862] Error Code 4: Internal Error ((Unnamed Layer* 46) [ElementWise]_output: tensor volume exceeds (2^31)-1, dimensions are [65536,13107,256]). Could you please explain what do you mean by try something between the default 224x224 and this large one?

5 replies

udasinnayan Aug 22, 2023
Author

Thank you so much @blaz-r, I tried with 1024*1024, which works fine. The problem occurred for huge dimensionality. @samet-akcay I need some help regarding this, in the anomalib components section I there is an script named PCA for dimensionality reduction. Could you please help me out to in patchcore model -> lightning_model.py, is it possible to use PCA for dimensionality reduction? Because the only problem I encountered is exceeds the tensor volume, and my custom image dimension is [65536,13107,256] if I use actual custom image, is it possible to use PCA to reduce the dimensionality here by any chance? Thanks in advance.

blaz-r Aug 22, 2023

What you are referring to as "custom image dimension of [65536,13107,256]" is not really an image dimension. I think it is the size of the distance matrix between embeddings. I believe that PCA can't really be used to reduce this. By setting resolution to 1024x1024 you are already performing a kind of dimensionality reduction by image downscaling.

What is the reason for such a large resolution if I may ask? Is the performance of the model with 1024x1024 not good? One thing you might try, since your original resolution is 2048x8192, is to keep this width-height ratio and downscale to something like 256x1024 or lets say 512x2048.

udasinnayan Aug 22, 2023
Author

@blaz-r Thanks for pointing out, and yes it's actually distance matrix. As you mentioned, It's stable with image size 256x1024 and 512x2048 if I downscale the image. The actual image is generated from other parts and can't be changed that's why I am looking for a way to fit the actual image size with this pipeline.

blaz-r Aug 22, 2023

The image is downscaled as part of the procedure. If you want your output to be original size, you could upscale the prediction back to 2048x8192.

Answer selected by udasinnayan

samet-akcay · 2023-08-22T09:43:50Z

samet-akcay
Aug 22, 2023
Maintainer

Let's convert this to a Q&A in discussions. We could continue from there. Thanks!

0 replies

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

ONNX export failed with custom image size trained with patchcore model #1290

Uh oh!

{{title}}

Uh oh!

Replies: 3 comments 5 replies

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Select a reply

Uh oh!

ONNX export failed with custom image size trained with patchcore model #1290

Uh oh!

udasinnayan Aug 21, 2023

Describe the bug

Dataset

Model

Steps to reproduce the behavior

OS information

Expected behavior

Screenshots

Pip/GitHub

What version/branch did you use?

Configuration YAML

Logs

Code of Conduct

Replies: 3 comments · 5 replies

Uh oh!

blaz-r Aug 21, 2023

Uh oh!

udasinnayan Aug 22, 2023 Author

Uh oh!

udasinnayan Aug 22, 2023 Author

Uh oh!

blaz-r Aug 22, 2023

Uh oh!

udasinnayan Aug 22, 2023 Author

Uh oh!

blaz-r Aug 22, 2023

Uh oh!

samet-akcay Aug 22, 2023 Maintainer

udasinnayan
Aug 21, 2023

Replies: 3 comments 5 replies

blaz-r
Aug 21, 2023

udasinnayan
Aug 22, 2023
Author

udasinnayan Aug 22, 2023
Author

udasinnayan Aug 22, 2023
Author

samet-akcay
Aug 22, 2023
Maintainer