Support online cls model prediction

tools/infer/text/README.md

@@ -238,6 +238,56 @@ Evaluation of the text spotting inference results on Ascend 910 with MindSpore 2
 2. Unless extra inidication, all experiments are run with `--det_limit_type`="min" and `--det_limit_side`=720.
 3. SVTR is run in mixed precision mode (amp_level=O2) since it is optimized for O2.
 
+## Text Direction Classification
+
+To run a text Angle classification on an input image or a directory that contains multiple images, please execute
+```shell
+python tools/infer/text/predict_cls.py  --image_dir {path_to_img or dir_to_imgs} --rec_algorithm MV3
+```
+When run, inference results are saved in `{args.draw_img_save_dir}/cls_results.txt`, where `--draw_img_save_dir` is the directory to save the results, which is the default setting for `./inference_results`. Here are some examples of the results.
+
+- Text Angle Classification
+
+<p align="center">
+  <img src="https://raw.githubusercontent.com/zhangjunlongtech/Material/refs/heads/main/CRNN_t1.png" width=150 />
+</p>
+<p align="center">
+  <em> word_01.png </em>
+</p>
+
+<p align="center">
+  <img src="https://raw.githubusercontent.com/zhangjunlongtech/Material/refs/heads/main/CRNN_t2.png" width=150 />
+</p>
+<p align="center">
+  <em> word_02.png </em>
+</p>
+
+Classification results:
+```text
+word_01.png   0
+word_02.png   180
+```
+
+**Note:**
+- for more parameters and usage, please run ` python tools/infer/text/predict_cls.py -h ` or view ` tools/infer/text/config.py `
+- Supports batch text Angle classification and single mode text Angle classification. Batch mode is enabled by default for speed. You can set the batch size with `--cls_batch_num`. You can also run in single image mode by setting `--cls_batch_mode` False.
+- You can set the specified weight file by setting `--cls_model_dir`, or ignore this setting. The system will load the default weight file.
+- The currently supported Angle classification network is mobilenet_v3. You can specify the model by setting `--cls_algorithm`, whose corresponding `--cls_amp_level` currently supports `O0`.
+
+### Supports angle classification algorithms and networks
+
+<center>
+
+  |**Algorithm name**|**Network name**|**Language**|
+  | :------: | :------: | :------: |
+  | MV3 | mobilenet_v3 | CH/EN|
+
+</center>
+
+The algorithm network is defined in `tools/infer/text/predict_cls.py`
+
+Currently, the listed models support classification of 0 and 180 degrees. Other angles of classification will be supported soon.
+
 ## Argument List
 
 All CLI argument definition can be viewed via `python tools/infer/text/predict_system.py -h` or reading `tools/infer/text/config.py`.

tools/infer/text/README_CN.md

@@ -230,6 +230,56 @@ python deploy/eval_utils/eval_pipeline.py --gt_path path/to/gt.txt --pred_path p
 
 3、SVTR在混合精度模式下运行（amp_level=O2），因为它针对O2进行了优化。
 
+## 文本方向分类器
+
+要对输入图像或包含多个图像的目录运行文本角度分类，请执行
+```shell
+python tools/infer/text/predict_cls.py  --image_dir {path_to_img or dir_to_imgs} --rec_algorithm MV3
+```
+运行后，推理结果保存在`{args.draw_img_save_dir}/cls_results.txt`中，其中`--draw_img_save_dir`是保存结果的目录，这是`./inference_results`的默认设置。下面是一些结果的例子。
+
+- 文本角度分类
+
+<p align="center">
+  <img src="https://raw.githubusercontent.com/zhangjunlongtech/Material/refs/heads/main/CRNN_t1.png" width=150 />
+</p>
+<p align="center">
+  <em> word_01.png </em>
+</p>
+
+<p align="center">
+  <img src="https://raw.githubusercontent.com/zhangjunlongtech/Material/refs/heads/main/CRNN_t2.png" width=150 />
+</p>
+<p align="center">
+  <em> word_02.png </em>
+</p>
+
+分类结果：
+```text
+word_01.png   0
+word_02.png   180
+```
+
+**注意事项：**
+- 有关更多参数说明和用法，请运行`python tools/infer/text/predict_cls.py -h`或查看`tools/infer/text/config.py`
+- 支持批量文本角度分类和单模文本角度分类。默认情况下启用批处理模式以提高速度。您可以通过`--cls_batch_num`设置批量大小。您还可以通过设置`--cls_batch_mode` False在单一图像模式下运行。
+- 可通过设置`--cls_model_dir` 来设置指定的权重文件，也可忽略此项设置，系统将加载默认权重文件。
+- 当前支持的角度分类器的网络为mobilenet_v3，可通过设置`--cls_algorithm` 来指定模型，其对应的`cls_amp_level`当前支持O0。
+
+### 支持的角度分类算法和网络
+
+<center>
+
+  |**算法名称**|**网络名称**|**语言**|
+  | :------: | :------: | :------: |
+  | MV3 | mobilenet_v3 | 中/英|
+
+</center>
+
+算法网络在`tools/infer/text/predict_cls.py`中定义
+
+目前，所列型号支持识别0与180度。对于其他角度的识别我们将很快予以支持。
+
 ## 参数列表
 
 所有CLI参数定义都可以通过`python tools/infer/text/predict_system.py -h`或`tools/infer/text/config.py`查看。

tools/infer/text/config.py

@@ -197,6 +197,36 @@ def create_parser():
         "--table_max_len", type=int, default=480, help="max length of the input image for table structure recognition."
     )
 
+
+    parser.add_argument(
+        "--cls_algorithm",
+        type=str,
+        default="MV3",
+        choices=["MV3"],
+        help="classification algorithm",
+    )
+    parser.add_argument(
+        "--cls_amp_level",
+        type=str,
+        default="O0",
+        choices=["O0", "O1", "O2", "O3"],
+        help="Auto Mixed Precision level. This setting only works on GPU and Ascend",
+    )
+    parser.add_argument(
+        "--cls_model_dir",
+        type=str,
+        help="directory containing the classification model checkpoint best.ckpt"
+        "or path to a specific checkpoint file.",
+    )
+    parser.add_argument(
+        "--cls_batch_mode",
+        type=str2bool,
+        default=True,
+        help="Whether to run classification inference in batch-mode, which is faster but may degrade the accuracy "
+        "due to padding or resizing to the same shape.",
+    )
+    parser.add_argument("--cls_batch_num", type=int, default=8)
+
     return parser
 
 

tools/infer/text/postprocess.py

@@ -75,7 +75,7 @@ def __init__(self, task="det", algo="DB", rec_char_dict_path=None, **kwargs):
                     character_dict_path=rec_char_dict_path,
                     use_space_char=False,
                 )
-
+                
             else:
                 raise ValueError(f"No postprocess config defined for {algo}. Please check the algorithm name.")
         elif task == "ser":
@@ -91,6 +91,8 @@ def __init__(self, task="det", algo="DB", rec_char_dict_path=None, **kwargs):
                 merge_no_span_structure=True,
                 box_shape="pad",
             )
+        elif task == "cls":
+            postproc_cfg = dict(name="ClsPostprocess", label_list=["0", "180"])
 
         postproc_cfg.update(kwargs)
         self.task = task
@@ -155,3 +157,6 @@ def __call__(self, pred, data=None, **kwargs):
         elif self.task == "table":
             output = self.postprocess(pred, labels=kwargs.get("labels"))
             return output
+        elif self.task == "cls":
+            output = self.postprocess(pred)
+            return output