【模型微调】基于ppv3检测模型训练微调，发现检测和原来的没啥区别

[hujiachun]

训练参数：

Global:
debug: false
use_gpu: true
epoch_num: 500
log_smooth_window: 20
print_batch_step: 10
save_model_dir: ./train_test/output/ch_PP-OCR_V3_det/
save_epoch_step: 100
eval_batch_step:

• 0
• 400
  cal_metric_during_train: false
  pretrained_model: /home/hll-qa/npl/PaddleOCR-release-2.7/train_test/npl_model/ch_PP-OCRv3_det_distill_train/student.pdparams
  checkpoints: null
  save_inference_dir: ./train_test/det_db_inference
  use_visualdl: false
  infer_img: doc/imgs_en/img_10.jpg
  save_res_path: ./checkpoints/det_db/predicts_db.txt
  distributed: true

Architecture:
model_type: det
algorithm: DB
Transform:
Backbone:
name: MobileNetV3
scale: 0.5
model_name: large
disable_se: True
Neck:
name: RSEFPN
out_channels: 96
shortcut: True
Head:
name: DBHead
k: 50

Loss:
name: DBLoss
balance_loss: true
main_loss_type: DiceLoss
alpha: 5
beta: 10
ohem_ratio: 3
Optimizer:
name: Adam
beta1: 0.9
beta2: 0.999
lr:
name: Cosine
learning_rate: 0.001
warmup_epoch: 2
regularizer:
name: L2
factor: 5.0e-05
PostProcess:
name: DBPostProcess
thresh: 0.3
box_thresh: 0.6
max_candidates: 1000
unclip_ratio: 1.5
Metric:
name: DetMetric
main_indicator: hmean
Train:
dataset:
name: SimpleDataSet
data_dir: ./train_test/npl_data_7_3
label_file_list:
- ./train_test/npl_data_7_3/Label.txt
ratio_list: [1.0]
transforms:
- DecodeImage:
img_mode: BGR
channel_first: false
- DetLabelEncode: null
- IaaAugment:
augmenter_args:
- type: Fliplr
args:
p: 0.5
- type: Affine
args:
rotate:
- -10
- 10
- type: Resize
args:
size:
- 0.5
- 3
- EastRandomCropData:
size:
- 960
- 960
max_tries: 50
keep_ratio: true
- MakeBorderMap:
shrink_ratio: 0.4
thresh_min: 0.3
thresh_max: 0.7
- MakeShrinkMap:
shrink_ratio: 0.4
min_text_size: 8
- NormalizeImage:
scale: 1./255.
mean:
- 0.485
- 0.456
- 0.406
std:
- 0.229
- 0.224
- 0.225
order: hwc
- ToCHWImage: null
- KeepKeys:
keep_keys:
- image
- threshold_map
- threshold_mask
- shrink_map
- shrink_mask
loader:
shuffle: true
drop_last: false
batch_size_per_card: 8
num_workers: 4
Eval:
dataset:
name: SimpleDataSet
data_dir: ./train_test/npl_data_7_3
label_file_list:
- ./train_test/npl_data_7_3/Label.txt
transforms:
- DecodeImage:
img_mode: BGR
channel_first: false
- DetLabelEncode: null
- DetResizeForTest: null
- NormalizeImage:
scale: 1./255.
mean:
- 0.485
- 0.456
- 0.406
std:
- 0.229
- 0.224
- 0.225
order: hwc
- ToCHWImage: null
- KeepKeys:
keep_keys:
- image
- shape
- polys
- ignore_tags
loader:
shuffle: false
drop_last: false
batch_size_per_card: 1
num_workers: 2

训练结果

[2024/07/03 13:16:39] ppocr INFO: save model in ./train_test/output/ch_PP-OCR_V3_det/iter_epoch_500
[2024/07/03 13:16:39] ppocr INFO: best metric, hmean: 0.9615975422427034, is_float16: False, precision: 0.9581632653061225, recall: 0.9650565262076053, fps: 124.01074288458881, best_epoch: 115
实测模型检测和原来训练之前检测结果无区别
虽然只放了50张图片，是这个原因吗？

[GreatV]

数据量有点少，参考 https://github.com/PaddlePaddle/PaddleOCR/blob/main/doc/doc_ch/FAQ.md#23-%E6%95%B0%E6%8D%AE%E9%87%8F%E8%AF%B4%E6%98%8E