Skip to content

Commit 3a42d50

Browse files
Bruce-pyBruce-py
authored
Delete det FCENet and update det 910* results. (#796)
* delete det FCENet and update det 910* results. * delete det FCENet and update det 910* results. * Update det README. * Update det README. * Update det README.
1 parent ab5e13f commit 3a42d50

33 files changed

+151
-2339
lines changed

‎README.md‎

Lines changed: 0 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -247,7 +247,6 @@ You can do MindSpore Lite inference in MindOCR using **MindOCR models** or **Thi
247247
- [x] [DBNet++](configs/det/dbnet/README.md) (TPAMI'2022)
248248
- [x] [PSENet](configs/det/psenet/README.md) (CVPR'2019)
249249
- [x] [EAST](configs/det/east/README.md)(CVPR'2017)
250-
- [x] [FCENet](configs/det/fcenet/README.md) (CVPR'2021)
251250

252251
</details>
253252

‎README_CN.md‎

Lines changed: 0 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -251,7 +251,6 @@ python tools/infer/text/predict_system.py --image_dir {path_to_img or dir_to_img
251251
- [x] [DBNet++](configs/det/dbnet/README_CN.md) (TPAMI'2022)
252252
- [x] [PSENet](configs/det/psenet/README_CN.md) (CVPR'2019)
253253
- [x] [EAST](configs/det/east/README_CN.md)(CVPR'2017)
254-
- [x] [FCENet](configs/det/fcenet/README_CN.md) (CVPR'2021)
255254
</details>
256255

257256
<details open markdown>

‎configs/det/dbnet/README.md‎

Lines changed: 9 additions & 26 deletions
Original file line numberDiff line numberDiff line change
@@ -299,29 +299,12 @@ python tools/eval.py -c=configs/det/dbnet/db_r50_icdar15.yaml
299299

300300
## Performance
301301

302-
### General Purpose Models
303-
304-
Here we present general purpose models that were trained on wide variety of tasks (real-world photos, street views, documents, etc.) and challenges (straight texts, curved texts, long text lines, etc.) with two primary languages: Chinese and English. These models can be used right off-the-shelf in your applications or for initialization of your models.
305-
306-
The models were trained on 12 public datasets (CTW, LSVT, RCTW-17, TextOCR, etc.) that contain wide range of images. The training set has 153,511 images and the validation set has 9,786 images.<br/>
307-
The test set consists of 598 images manually selected from the above-mentioned datasets.
302+
DBNet and DBNet++ were trained on the ICDAR2015, MSRA-TD500, SCUT-CTW1500, Total-Text, and MLT2017 datasets. In addition, we conducted pre-training on the ImageNet or SynthText dataset and provided a URL to download pretrained weights. All training results are as follows:
308303

309304
Experiments are tested on ascend 910* with mindspore 2.3.1 graph mode.
310305

311-
*coming soon*
312-
313-
Experiments are tested on ascend 910 with mindspore 2.3.1 graph mode.
314-
315-
*coming soon*
316-
317-
318-
### Specific Purpose Models
319-
320-
DBNet and DBNet++ were trained on the ICDAR2015, MSRA-TD500, SCUT-CTW1500, Total-Text, and MLT2017 datasets. In addition, we conducted pre-training on the SynthText dataset and provided a URL to download pretrained weights. All training results are as follows:
321-
322306
#### ICDAR2015
323307

324-
Experiments are tested on ascend 910* with mindspore 2.3.1 graph mode.
325308

326309
| **model name** | **backbone** | **pretrained** | **cards** | **batch size** | **jit level** | **graph compile** | **ms/step** | **img/s** | **recall** | **precision** | **f-score** | **recipe** | **weight** |
327310
| :------------: | :----------: | :------------: | :-------: | :------------: | :-----------: | :---------------: | :---------: | :-------: | :--------: | :-----------: | :---------: | :------------------------------------: | :--------------------------------------------------------------------------------------------------------: |
@@ -336,14 +319,14 @@ Experiments are tested on ascend 910 with mindspore 2.3.1 graph mode.
336319

337320
#### ICDAR2015
338321

339-
| **model name** | **backbone** | **pretrained** | **cards** | **batch size** | **jit level** | **graph compile** | **ms/step** | img/s | **recall** | **precision** | **f-score** | **recipe** | **weight** |
340-
| :------------: | :----------: | :------------: | :-------: | :------------: | :-----------: | :---------------: | :---------: | :---: | :--------: | :-----------: | :---------: | :------------------------------------: | :-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
341-
| DBNet | MobileNetV3 | ImageNet | 1 | 10 | O2 | 321.15 s | 100 | 100 | 76.31% | 78.27% | 77.28% | [yaml](db_mobilenetv3_icdar15.yaml) | [ckpt](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnet_mobilenetv3-62c44539.ckpt) \| [mindir](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnet_mobilenetv3-62c44539-f14c6a13.mindir) |
342-
| DBNet | MobileNetV3 | ImageNet | 8 | 8 | O2 | 309.39 s | 66.64 | 960 | 76.22% | 77.98% | 77.09% | [yaml](db_mobilenetv3_icdar15_8p.yaml) | Coming soon |
343-
| DBNet | ResNet-18 | ImageNet | 1 | 20 | O2 | 75.23 s | 185.19 | 108 | 80.12% | 83.41% | 81.73% | [yaml](db_r18_icdar15.yaml) | [ckpt](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnet_resnet18-0c0c4cfa.ckpt) \| [mindir](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnet_resnet18-0c0c4cfa-cf46eb8b.mindir) |
344-
| DBNet | ResNet-50 | ImageNet | 1 | 10 | O2 | 110.54 s | 132.98 | 75.2 | 83.53% | 86.62% | 85.05% | [yaml](db_r50_icdar15.yaml) | [ckpt](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnet_resnet50-c3a4aa24.ckpt) \| [mindir](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnet_resnet50-c3a4aa24-fbf95c82.mindir) |
345-
| DBNet | ResNet-50 | ImageNet | 8 | 10 | O2 | 107.91 s | 183.92 | 435 | 82.62% | 88.54% | 85.48% | [yaml](db_r50_icdar15_8p.yaml) | Coming soon |
346-
| DBNet++ | ResNet-50 | SynthText | 1 | 32 | O2 | 184.74 s | 409.21 | 78.2 | 86.81% | 86.85% | 86.86% | [yaml](dbpp_r50_icdar15_910.yaml) | [ckpt](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnetpp_resnet50_910-35dc71f2.ckpt) \| [mindir](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnetpp_resnet50_910-35dc71f2-e61a9c37.mindir) |
322+
| **model name** | **backbone** | **pretrained** | **cards** | **batch size** | **jit level** | **graph compile** | **ms/step** | **img/s** | **recall** | **precision** | **f-score** | **recipe** | **weight** |
323+
| :------------: | :----------: | :------------: | :-------: | :------------: | :-----------: | :---------------: | :---------: |:---------:| :--------: | :-----------: | :---------: | :------------------------------------: | :-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
324+
| DBNet | MobileNetV3 | ImageNet | 1 | 10 | O2 | 321.15 s | 100 | 100 | 76.31% | 78.27% | 77.28% | [yaml](db_mobilenetv3_icdar15.yaml) | [ckpt](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnet_mobilenetv3-62c44539.ckpt) \| [mindir](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnet_mobilenetv3-62c44539-f14c6a13.mindir) |
325+
| DBNet | MobileNetV3 | ImageNet | 8 | 8 | O2 | 309.39 s | 66.64 | 960 | 76.22% | 77.98% | 77.09% | [yaml](db_mobilenetv3_icdar15_8p.yaml) | Coming soon |
326+
| DBNet | ResNet-18 | ImageNet | 1 | 20 | O2 | 75.23 s | 185.19 | 108 | 80.12% | 83.41% | 81.73% | [yaml](db_r18_icdar15.yaml) | [ckpt](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnet_resnet18-0c0c4cfa.ckpt) \| [mindir](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnet_resnet18-0c0c4cfa-cf46eb8b.mindir) |
327+
| DBNet | ResNet-50 | ImageNet | 1 | 10 | O2 | 110.54 s | 132.98 | 75.2 | 83.53% | 86.62% | 85.05% | [yaml](db_r50_icdar15.yaml) | [ckpt](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnet_resnet50-c3a4aa24.ckpt) \| [mindir](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnet_resnet50-c3a4aa24-fbf95c82.mindir) |
328+
| DBNet | ResNet-50 | ImageNet | 8 | 10 | O2 | 107.91 s | 183.92 | 435 | 82.62% | 88.54% | 85.48% | [yaml](db_r50_icdar15_8p.yaml) | Coming soon |
329+
| DBNet++ | ResNet-50 | SynthText | 1 | 32 | O2 | 184.74 s | 409.21 | 78.2 | 86.81% | 86.85% | 86.86% | [yaml](dbpp_r50_icdar15_910.yaml) | [ckpt](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnetpp_resnet50_910-35dc71f2.ckpt) \| [mindir](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnetpp_resnet50_910-35dc71f2-e61a9c37.mindir) |
347330

348331
> The input_shape for exported DBNet MindIR and DBNet++ MindIR in the links are `(1,3,736,1280)` and `(1,3,1152,2048)`, respectively.
349332

‎configs/det/dbnet/README_CN.md‎

Lines changed: 9 additions & 27 deletions
Original file line numberDiff line numberDiff line change
@@ -279,25 +279,7 @@ python tools/eval.py --config configs/det/dbnet/db_r50_icdar15.yaml
279279

280280
## 性能表现
281281

282-
### 通用泛化模型
283-
284-
本节提供了一些通过泛化模型，该模型使用中文和英文两种语言训练，针对各种不同的任务和挑战，包括真实世界图片，街景图片，文档，弯曲文本，长文本等。这些模型可直接用于下游任务，也可直接作为预训练权重。
285-
286-
这些模型在12个公开数据集上训练，包括CTW，LSVT，RCTW-17，TextOCR等，其中训练集包含153511张图片，验证集包含9786张图片。<br/>
287-
从上述数据集中手动选择598张未被训练集和验证集使用的图片构成测试集。
288-
289-
在采用图模式的ascend 910*上实验结果，mindspore版本为2.3.1
290-
291-
*即将到来*
292-
293-
在采用图模式的ascend 910上实验结果，mindspore版本为2.3.1
294-
295-
*即将到来*
296-
297-
298-
### 细分领域模型
299-
300-
DBNet和DBNet++在ICDAR2015，MSRA-TD500，SCUT-CTW1500，Total-Text和MLT2017数据集上训练。另外，我们在SynthText数据集上进行了预训练，并提供预训练权重下载链接。所有训练结果如下：
282+
DBNet和DBNet++在ICDAR2015，MSRA-TD500，SCUT-CTW1500，Total-Text和MLT2017数据集上训练。另外，我们在ImageNet和SynthText数据集上进行了预训练，并提供预训练权重下载链接。所有训练结果如下：
301283

302284
在采用图模式的ascend 910*上实验结果，mindspore版本为2.3.1
303285

@@ -318,14 +300,14 @@ DBNet和DBNet++在ICDAR2015，MSRA-TD500，SCUT-CTW1500，Total-Text和MLT2017
318300

319301
#### ICDAR2015
320302

321-
| **model name** | **backbone** | **pretrained** | **cards** | **batch size** | **jit level** | **graph compile** | **ms/step** | img/s | **recall** | **precision** | **f-score** | **recipe** | **weight** |
322-
| :------------: | :----------: | :------------: | :-------: | :------------: | :-----------: | :---------------: | :---------: | :---: | :--------: | :-----------: | :---------: | :------------------------------------: | :-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
323-
| DBNet | MobileNetV3 | ImageNet | 1 | 10 | O2 | 321.15 s | 100 | 100 | 76.31% | 78.27% | 77.28% | [yaml](db_mobilenetv3_icdar15.yaml) | [ckpt](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnet_mobilenetv3-62c44539.ckpt) \| [mindir](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnet_mobilenetv3-62c44539-f14c6a13.mindir) |
324-
| DBNet | MobileNetV3 | ImageNet | 8 | 8 | O2 | 309.39 s | 66.64 | 960 | 76.22% | 77.98% | 77.09% | [yaml](db_mobilenetv3_icdar15_8p.yaml) | Coming soon |
325-
| DBNet | ResNet-18 | ImageNet | 1 | 20 | O2 | 75.23 s | 185.19 | 108 | 80.12% | 83.41% | 81.73% | [yaml](db_r18_icdar15.yaml) | [ckpt](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnet_resnet18-0c0c4cfa.ckpt) \| [mindir](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnet_resnet18-0c0c4cfa-cf46eb8b.mindir) |
326-
| DBNet | ResNet-50 | ImageNet | 1 | 10 | O2 | 110.54 s | 132.98 | 75.2 | 83.53% | 86.62% | 85.05% | [yaml](db_r50_icdar15.yaml) | [ckpt](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnet_resnet50-c3a4aa24.ckpt) \| [mindir](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnet_resnet50-c3a4aa24-fbf95c82.mindir) |
327-
| DBNet | ResNet-50 | ImageNet | 8 | 10 | O2 | 107.91 s | 183.92 | 435 | 82.62% | 88.54% | 85.48% | [yaml](db_r50_icdar15_8p.yaml) | Coming soon |
328-
| DBNet++ | ResNet-50 | SynthText | 1 | 32 | O2 | 184.74 s | 409.21 | 78.2 | 86.81% | 86.85% | 86.86% | [yaml](dbpp_r50_icdar15_910.yaml) | [ckpt](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnetpp_resnet50_910-35dc71f2.ckpt) \| [mindir](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnetpp_resnet50_910-35dc71f2-e61a9c37.mindir) |
303+
| **model name** | **backbone** | **pretrained** | **cards** | **batch size** | **jit level** | **graph compile** | **ms/step** | **img/s** | **recall** | **precision** | **f-score** | **recipe** | **weight** |
304+
| :------------: | :----------: | :------------: | :-------: | :------------: | :-----------: | :---------------: | :---------: |:----------:| :--------: | :-----------: | :---------: | :------------------------------------: | :-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
305+
| DBNet | MobileNetV3 | ImageNet | 1 | 10 | O2 | 321.15 s | 100 | 100 | 76.31% | 78.27% | 77.28% | [yaml](db_mobilenetv3_icdar15.yaml) | [ckpt](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnet_mobilenetv3-62c44539.ckpt) \| [mindir](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnet_mobilenetv3-62c44539-f14c6a13.mindir) |
306+
| DBNet | MobileNetV3 | ImageNet | 8 | 8 | O2 | 309.39 s | 66.64 | 960 | 76.22% | 77.98% | 77.09% | [yaml](db_mobilenetv3_icdar15_8p.yaml) | Coming soon |
307+
| DBNet | ResNet-18 | ImageNet | 1 | 20 | O2 | 75.23 s | 185.19 | 108 | 80.12% | 83.41% | 81.73% | [yaml](db_r18_icdar15.yaml) | [ckpt](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnet_resnet18-0c0c4cfa.ckpt) \| [mindir](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnet_resnet18-0c0c4cfa-cf46eb8b.mindir) |
308+
| DBNet | ResNet-50 | ImageNet | 1 | 10 | O2 | 110.54 s | 132.98 | 75.2 | 83.53% | 86.62% | 85.05% | [yaml](db_r50_icdar15.yaml) | [ckpt](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnet_resnet50-c3a4aa24.ckpt) \| [mindir](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnet_resnet50-c3a4aa24-fbf95c82.mindir) |
309+
| DBNet | ResNet-50 | ImageNet | 8 | 10 | O2 | 107.91 s | 183.92 | 435 | 82.62% | 88.54% | 85.48% | [yaml](db_r50_icdar15_8p.yaml) | Coming soon |
310+
| DBNet++ | ResNet-50 | SynthText | 1 | 32 | O2 | 184.74 s | 409.21 | 78.2 | 86.81% | 86.85% | 86.86% | [yaml](dbpp_r50_icdar15_910.yaml) | [ckpt](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnetpp_resnet50_910-35dc71f2.ckpt) \| [mindir](https://download.mindspore.cn/toolkits/mindocr/dbnet/dbnetpp_resnet50_910-35dc71f2-e61a9c37.mindir) |
329311

330312

331313
> 链接中模型DBNet的MindIR导出时的输入Shape为`(1,3,736,1280)`，模型DBNet++的MindIR导出时的输入Shape为`(1,3,1152,2048)`。

0 commit comments

Comments
 (0)