- add author loss, The results are compared as follows
| Method | Precision (%) | Recall (%) | F-measure (%) | fps |
|---|---|---|---|---|
| PSENet-1s with resnet152 batch 8 myloss | 85.47 | 78.76 | 81.98 | 1.49 |
| PSENet-1s with resnet152 batch 8 author loss | 85.04 | 79.68 | 82.27 | 1.48 |
resnet50 and resnet152: bauduyun extract code: rxjf
- config the
trainroot,testrootin config.py - use fellow script to run
python3 train.pyeval.py is used to test model on test dataset
- config
model_path,data_path,gt_path,save_pathin eval.py - use fellow script to test
python3 eval.pypredict.py is used to inference on single image
- config
model_path,img_path,gt_path,save_pathin predict.py - use fellow script to predict
python3 predict.pyThe project is still under development.
only train with ICDAR2015 dataset with single 1080ti
my implementation with author loss use adam and MultiStepLR
| Method | Precision (%) | Recall (%) | F-measure (%) | fps |
|---|---|---|---|---|
| PSENet-1s with resnet50 batch 8 | 83.49 | 79.62 | 81.51 | 1.76 |
| PSENet-2s with resnet50 batch 8 | 83.37 | 79.68 | 81.48 | 3.55 |
| PSENet-4s with resnet50 batch 8 | 82.44 | 78.91 | 80.63 | 4.43 |
| PSENet-1s with resnet152 batch 4 | 85.04 | 79.68 | 82.27 | 1.48 |
| PSENet-2s with resnet152 batch 4 | 84.88 | 79.20 | 81.94 | 2.56 |
| PSENet-4s with resnet152 batch 4 | 83.81 | 78.76 | 81.21 | 2.99 |
my implementation with my loss use adam and warm_up
| Method | Precision (%) | Recall (%) | F-measure (%) | fps |
|---|---|---|---|---|
| PSENet-1s with resnet50 batch 8 | tbd | tbd | tbd | 1.76 |
| PSENet-2s with resnet50 batch 8 | tbd | tbd | tbd | 3.55 |
| PSENet-4s with resnet50 batch 8 | tbd | tbd | tbd | 4.43 |
| PSENet-1s with resnet152 batch 4 | 85.24 | 80.11 | 82.60 | 1.48 |
| PSENet-2s with resnet152 batch 4 | 85.02 | 80.06 | 82.46 | 2.56 |
| PSENet-4s with resnet152 batch 4 | 83.46 | 79.00 | 81.17 | 2.99 |
official implementation use SGD and StepLR
| Method | Precision (%) | Recall (%) | F-measure (%) | fps |
|---|---|---|---|---|
| PSENet-1s with resnet50 batch 8 | 84.15 | 80.26 | 82.16 | 1.76 |
| PSENet-2s with resnet50 batch 8 | 83.61 | 79.82 | 81.67 | 3.72 |
| PSENet-4s with resnet50 batch 8 | 81.90 | 78.23 | 80.03 | 4.51 |
| PSENet-1s with resnet152 batch 4 | 82.87 | 78.76 | 80.77 | 1.53 |
| PSENet-2s with resnet152 batch 4 | 82.33 | 78.33 | 80.28 | 2.61 |
| PSENet-4s with resnet152 batch 4 | 81.19 | 77.13 | 79.11 | 3.00 |
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent