custom data training instructions solov2

[sbr300]

1. If is feasible to train on custom data kindly provide the instructions.
2. if you feel registering the data with detection instructions is faceable. kindly provide how to pass the registered data through train.py

[WXinlong]

Please follow the default instructions in detectron2 to train on custom data. There should be no special setting for SOLOv2 when training on your custom data.

[EllisHui]

Traceback (most recent call last): File "/home/huiqiang/code/AdelaiDet/adet/modeling/solov2/train_my_custom_net_5.py", line 197, in args=(args,), File "/home/huiqiang/code/AdelaiDet/detectron2/detectron2/engine/launch.py", line 87, in launch main_func(*args) File "/home/huiqiang/code/AdelaiDet/adet/modeling/solov2/train_my_custom_net_5.py", line 178, in main trainer = Trainer(cfg) File "/home/huiqiang/code/AdelaiDet/detectron2/detectron2/engine/defaults.py", line 377, in init model = self.build_model(cfg) File "/home/huiqiang/code/AdelaiDet/detectron2/detectron2/engine/defaults.py", line 506, in build_model model = build_model(cfg) File "/home/huiqiang/code/AdelaiDet/detectron2/detectron2/modeling/meta_arch/build.py", line 27, in build_model model = META_ARCH_REGISTRY.get(meta_arch)(cfg) File "/home/huiqiang/code/AdelaiDet/adet/modeling/solov2/solov2.py", line 42, in init self.scale_ranges = cfg.MODEL.SOLOV2.FPN_SCALE_RANGES File "/home/huiqiang/Envs/pytorch_hq_gpus_solov2OfAdelaiDet_detectron2/lib/python3.7/site-packages/yacs/config.py", line 141, in getattr raise AttributeError(name) AttributeError: SOLOV2

[sbr300]

Please follow the default instructions in detectron2 to train on custom data. There should be no special setting for SOLOv2 when training on your custom data.

I requested the steps because it doesn't work.