Not clear how to prepare custom data for training

Hello community!

I have a question regarding the preparation of data to train this network with my own data. Right now I have two folders (images and labels) with the same name per file. The labels are in xml (VOC). I don’t understand very well the training manual and I don’t know how to continue. What exactly do I have to change in the exps/example/custom/yolox_s.py?

A question I have at this point is, what exactly are these files?

self.train_ann = "instances_train2017.json".
self.val_ann = "instances_val2017.json"

Best regards

[hongsamvo]

In case your dataset in VOC format, you can follow the author’s guide, remember to change your dataset name train and val, in voc.py here, and make sure no error at here, change your classes, remember to link dataset as guide ln -s /path/to/your/VOCdevkit ./datasets/VOCdevkit and follow python tools/train.py -f exps/example/yolox_voc/yolox_voc_s.py -d 8 -b 64 --fp16 -o -c /path/to/yolox_s.pth [--cache]

`

[hongsamvo]

Yes, actually the error will happen here and other similar places, replace the term "VOC" + year by year, and one error here, ignore if else, choose metric you want.