Zhongdao / Towards-Realtime-MOT

Joint Detection and Embedding for fast multi-object tracking
MIT License
2.37k stars 539 forks source link

training and inference preprocessing (255.0 scaling) #227

Open ZiyanZhu1994 opened 3 years ago

ZiyanZhu1994 commented 3 years ago

Thanks for the excellent work!!! I have a question regrading the scaling in preprocess.

During inference time, the image is scale to values between 0 and 1.

https://github.com/Zhongdao/Towards-Realtime-MOT/blob/88df0e5615b90877c3f5e030416d9dc0cee12cdc/utils/datasets.py#L52

However, during the training time, I don't find any scaling on image. Is that scaling not important for training?

Many thanks!