文字检测模型，模型不训练，也不报错。 #12299

kikiyaYang · 2024-03-29T08:16:30Z

kikiyaYang
Mar 29, 2024

以下是我的训练过程，他卡在'ppocr INFO: During the training process, after the 0th iteration, an evaluation is run every 20 iterations'，不再训练。
$ python3 tools/train.py -c configs/det/det_r18_vd_ct_test.yml -o Optimizer.base_lr=0.0001
[2024/03/29 16:06:06] ppocr INFO: Architecture :
[2024/03/29 16:06:06] ppocr INFO: Backbone :
[2024/03/29 16:06:06] ppocr INFO: layers : 18
[2024/03/29 16:06:06] ppocr INFO: name : ResNet_vd
[2024/03/29 16:06:06] ppocr INFO: Head :
[2024/03/29 16:06:06] ppocr INFO: hidden_dim : 128
[2024/03/29 16:06:06] ppocr INFO: in_channels : 512
[2024/03/29 16:06:06] ppocr INFO: name : CT_Head
[2024/03/29 16:06:06] ppocr INFO: num_classes : 3
[2024/03/29 16:06:06] ppocr INFO: Neck :
[2024/03/29 16:06:06] ppocr INFO: name : CTFPN
[2024/03/29 16:06:06] ppocr INFO: Transform : None
[2024/03/29 16:06:06] ppocr INFO: algorithm : CT
[2024/03/29 16:06:06] ppocr INFO: model_type : det
[2024/03/29 16:06:06] ppocr INFO: Eval :
[2024/03/29 16:06:06] ppocr INFO: dataset :
[2024/03/29 16:06:06] ppocr INFO: data_dir : ./train_data/train/
[2024/03/29 16:06:06] ppocr INFO: label_file_list : ['./train_data/xml2train_label.txt']
[2024/03/29 16:06:06] ppocr INFO: name : SimpleDataSet
[2024/03/29 16:06:06] ppocr INFO: ratio_list : [1.0]
[2024/03/29 16:06:06] ppocr INFO: transforms :
[2024/03/29 16:06:06] ppocr INFO: DecodeImage :
[2024/03/29 16:06:06] ppocr INFO: channel_first : False
[2024/03/29 16:06:06] ppocr INFO: img_mode : RGB
[2024/03/29 16:06:06] ppocr INFO: CTLabelEncode : None
[2024/03/29 16:06:06] ppocr INFO: ScaleAlignedShort : None
[2024/03/29 16:06:06] ppocr INFO: NormalizeImage :
[2024/03/29 16:06:06] ppocr INFO: order : hwc
[2024/03/29 16:06:06] ppocr INFO: ToCHWImage : None
[2024/03/29 16:06:06] ppocr INFO: KeepKeys :
[2024/03/29 16:06:06] ppocr INFO: keep_keys : ['image', 'shape', 'polys', 'texts']
[2024/03/29 16:06:06] ppocr INFO: loader :
[2024/03/29 16:06:06] ppocr INFO: batch_size_per_card : 1
[2024/03/29 16:06:06] ppocr INFO: drop_last : False
[2024/03/29 16:06:06] ppocr INFO: num_workers : 2
[2024/03/29 16:06:06] ppocr INFO: shuffle : False
[2024/03/29 16:06:06] ppocr INFO: Global :
[2024/03/29 16:06:06] ppocr INFO: cal_metric_during_train : False
[2024/03/29 16:06:06] ppocr INFO: checkpoints : None
[2024/03/29 16:06:06] ppocr INFO: distributed : False
[2024/03/29 16:06:06] ppocr INFO: epoch_num : 600
[2024/03/29 16:06:06] ppocr INFO: eval_batch_step : [0, 20]
[2024/03/29 16:06:06] ppocr INFO: infer_img : doc/imgs/test.jpg
[2024/03/29 16:06:06] ppocr INFO: log_smooth_window : 20
[2024/03/29 16:06:06] ppocr INFO: pretrained_model : ./pretrain_models/ResNet18_vd_pretrained.pdparams
[2024/03/29 16:06:06] ppocr INFO: print_batch_step : 10
[2024/03/29 16:06:06] ppocr INFO: save_epoch_step : 10
[2024/03/29 16:06:06] ppocr INFO: save_inference_dir : inference/test
[2024/03/29 16:06:06] ppocr INFO: save_model_dir : ./output/det_ct/
[2024/03/29 16:06:06] ppocr INFO: save_res_path : ./output/det_ct/predicts_ct.txt
[2024/03/29 16:06:06] ppocr INFO: use_gpu : True
[2024/03/29 16:06:06] ppocr INFO: use_visualdl : False
[2024/03/29 16:06:06] ppocr INFO: Loss :
[2024/03/29 16:06:06] ppocr INFO: name : CTLoss
[2024/03/29 16:06:06] ppocr INFO: Metric :
[2024/03/29 16:06:06] ppocr INFO: main_indicator : f_score
[2024/03/29 16:06:06] ppocr INFO: name : CTMetric
[2024/03/29 16:06:06] ppocr INFO: Optimizer :
[2024/03/29 16:06:06] ppocr INFO: base_lr : 0.0001
[2024/03/29 16:06:06] ppocr INFO: lr :
[2024/03/29 16:06:06] ppocr INFO: end_lr : 0.0
[2024/03/29 16:06:06] ppocr INFO: epochs : 600
[2024/03/29 16:06:06] ppocr INFO: learning_rate : 0.001
[2024/03/29 16:06:06] ppocr INFO: name : Linear
[2024/03/29 16:06:06] ppocr INFO: power : 0.9
[2024/03/29 16:06:06] ppocr INFO: step_each_epoch : 10
[2024/03/29 16:06:06] ppocr INFO: name : Adam
[2024/03/29 16:06:06] ppocr INFO: PostProcess :
[2024/03/29 16:06:06] ppocr INFO: box_type : poly
[2024/03/29 16:06:06] ppocr INFO: name : CTPostProcess
[2024/03/29 16:06:06] ppocr INFO: Train :
[2024/03/29 16:06:06] ppocr INFO: dataset :
[2024/03/29 16:06:06] ppocr INFO: data_dir : ./train_data/train/
[2024/03/29 16:06:06] ppocr INFO: label_file_list : ['./train_data/xml2train_label.txt']
[2024/03/29 16:06:06] ppocr INFO: name : SimpleDataSet
[2024/03/29 16:06:06] ppocr INFO: ratio_list : [1.0]
[2024/03/29 16:06:06] ppocr INFO: transforms :
[2024/03/29 16:06:06] ppocr INFO: DecodeImage :
[2024/03/29 16:06:06] ppocr INFO: channel_first : False
[2024/03/29 16:06:06] ppocr INFO: img_mode : RGB
[2024/03/29 16:06:06] ppocr INFO: CTLabelEncode : None
[2024/03/29 16:06:06] ppocr INFO: RandomScale : None
[2024/03/29 16:06:06] ppocr INFO: MakeShrink : None
[2024/03/29 16:06:06] ppocr INFO: GroupRandomHorizontalFlip : None
[2024/03/29 16:06:06] ppocr INFO: GroupRandomRotate : None
[2024/03/29 16:06:06] ppocr INFO: GroupRandomCropPadding : None
[2024/03/29 16:06:06] ppocr INFO: MakeCentripetalShift : None
[2024/03/29 16:06:06] ppocr INFO: ColorJitter :
[2024/03/29 16:06:06] ppocr INFO: brightness : 0.125
[2024/03/29 16:06:06] ppocr INFO: saturation : 0.5
[2024/03/29 16:06:06] ppocr INFO: ToCHWImage : None
[2024/03/29 16:06:06] ppocr INFO: NormalizeImage : None
[2024/03/29 16:06:06] ppocr INFO: KeepKeys :
[2024/03/29 16:06:06] ppocr INFO: keep_keys : ['image', 'gt_kernel', 'training_mask', 'gt_instance', 'gt_kernel_instance', 'training_mask_distance', 'gt_distance']
[2024/03/29 16:06:06] ppocr INFO: loader :
[2024/03/29 16:06:06] ppocr INFO: batch_size_per_card : 1
[2024/03/29 16:06:06] ppocr INFO: drop_last : True
[2024/03/29 16:06:06] ppocr INFO: num_workers : 2
[2024/03/29 16:06:06] ppocr INFO: shuffle : True
[2024/03/29 16:06:06] ppocr INFO: profiler_options : None
[2024/03/29 16:06:06] ppocr INFO: train with paddle 2.4.2 and device Place(gpu:0)
[2024/03/29 16:06:06] ppocr INFO: Initialize indexs of datasets:['./train_data/xml2train_label.txt']
list index out of range
[2024/03/29 16:06:06] ppocr INFO: Initialize indexs of datasets:['./train_data/xml2train_label.txt']
W0329 16:06:06.432868 18376 gpu_resources.cc:61] Please NOTE: device: 0, GPU Compute Capability: 8.6, Driver API Version: 11.8, Runtime API Version: 11.7
W0329 16:06:06.434868 18376 gpu_resources.cc:91] device: 0, cuDNN Version: 8.9.
[2024/03/29 16:06:07] ppocr INFO: train dataloader has 100 iters
[2024/03/29 16:06:07] ppocr INFO: valid dataloader has 100 iters
[2024/03/29 16:06:07] ppocr INFO: load pretrain successful from ./pretrain_models/ResNet18_vd_pretrained
now_debugg
[2024/03/29 16:06:07] ppocr INFO: During the training process, after the 0th iteration, an evaluation is run every 20 iterations

我的数据集一共100张图像和标注，望大佬解决！

Answered by kikiyaYang

Apr 2, 2024

更新了版本，解决了

View full answer

zhangyubo0722 · 2024-04-01T06:14:12Z

zhangyubo0722
Apr 1, 2024
Collaborator

后台gpu有利用率吗

0 replies

kikiyaYang · 2024-04-02T07:34:46Z

kikiyaYang
Apr 2, 2024
Author

更新了版本，解决了

0 replies

zyyyyy123lao · 2024-07-05T07:54:38Z

zyyyyy123lao
Jul 5, 2024

我也遇到了同样的问题，请问是更新了什么版本

0 replies

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

文字检测模型，模型不训练，也不报错。 #12299

Uh oh!

{{title}}

Uh oh!

Replies: 3 comments

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Select a reply

Uh oh!

文字检测模型，模型不训练，也不报错。 #12299

Uh oh!

kikiyaYang Mar 29, 2024

Replies: 3 comments

Uh oh!

zhangyubo0722 Apr 1, 2024 Collaborator

Uh oh!

kikiyaYang Apr 2, 2024 Author

Uh oh!

zyyyyy123lao Jul 5, 2024

kikiyaYang
Mar 29, 2024

zhangyubo0722
Apr 1, 2024
Collaborator

kikiyaYang
Apr 2, 2024
Author

zyyyyy123lao
Jul 5, 2024