How to detect characters from a certain image using trained models?

[renwoxing2016]

我正在看tutorial介绍及代码，希望能给与指导，利用已经训练的model，应该参考哪部分代码。

[yuantailing]

Refer to tutorial 3-detection,

1. skip `Training steps' and download trained models to corresponding file path
2. replace detection testing set with your image(s)
3. run `Predicting steps'

[renwoxing2016]

because skip training steps, I run predicting steps, but report 'assert os.path.exists(obj)' of exists_and_newer funciton, I guess that it has not the file of 'cates.json'. 'cates.json' can be download?

[yuantailing]

See #8 ,run python3 decide_cates.py with TRAIN+VAL.

[yuantailing]

Thanks for your good issue, tutorial is now updated, a new tip has been added:

If you are using trained models without training steps, you need to run python3 decide_cates.py (see 'Decide categories' section) with TRAIN+VAL to generate cates.json, the map from label ID to character category.