Add online prediction of text direction classification (CLS) model

[zhangjunlongtech]

Thank you for your contribution to the MindOCR repo.
Before submitting this PR, please make sure:

• [✔] You have read the Contributing Guidelines on pull requests
• [✔] Your code builds clean without any errors or warnings
• [✔] You are using approved terminology
• [✔] You have added unit tests

Motivation

Currently mobilenet_v3 classification (CLS) model only supports offline inference with mindspore lite. This PR is adding online mobilenet_v3 classification model inference for text angle classification (CLS) task. Add predict_cls.py for everyone to use.

Test Plan

Running the following command and check the output files under ./inferrence_results

python tools/infer/text/predict_cls.py  --image_dir {path_to_img or dir_to_imgs} --rec_algorithm MV3

We can also run in single image mode by setting --cls_batch_mode False.

python tools/infer/text/predict_cls.py  --image_dir {path_to_img or dir_to_imgs} --rec_algorithm MV3 --cls_batch_mode False

The target classification image is

The cls task output should looks like this

mindocr INFO - Init classification model: MV3 --> cls_mobilenet_v3_small_100_model. Model weights loaded from pretrained url
mindocr INFO - num images for cls: 1
mindocr INFO - CLS img idx range: [0, 1)
mindocr INFO - All cls res: [('180', 1.0)]
mindocr INFO - Done! Text angle classification results saved in ./inference_results
mindocr INFO - Time cost: 6.98498272895813, FPS: 0.14316427667805537