Open aswa123 opened 3 years ago
Please show me your error message so I can help you.
Traceback (most recent call last):
File "/home/swap/miniconda3/envs/Res2Net-maskrcnn-master/lib/python3.8/site-packages/IPython/core/interactiveshell.py", line 3427, in run_code
exec(code_obj, self.user_global_ns, self.user_ns)
File "
Do you use the following command to train the model? export NGPUS=8 python -m torch.distributed.launch --nproc_per_node=$NGPUS /path_to_maskrcnn_benchmark/tools/train_net.py --config-file "configs/pytorch_mask_rcnn_R2_50_s4_FPN_2x.yaml"
The train function is def train(cfg, local_rank, distributed):
And it should receive 3 arguments instead of 2. I didn't find this bug when running this repo.
How you install this repo first following https://github.com/Res2Net/Res2Net-maskrcnn/blob/master/INSTALL.md ?
python tools/train_net.py --config-file "configs/pytorch_mask_rcnn_R2_50_s4_FPN_2x.yaml" SOLVER.IMS_PER_BATCH 2 SOLVER.BASE_LR 0.0025 SOLVER.MAX_ITER 720000 SOLVER.STEPS "(480000, 640000)" TEST.IMS_PER_BATCH 1
Do you use the following command to train the model? export NGPUS=8 python -m torch.distributed.launch --nproc_per_node=$NGPUS /path_to_maskrcnn_benchmark/tools/train_net.py --config-file "configs/pytorch_mask_rcnn_R2_50_s4_FPN_2x.yaml"
How you install this repo first following https://github.com/Res2Net/Res2Net-maskrcnn/blob/master/INSTALL.md ?
yes.
You can try the distributed training mode. I didn't use the python tools/train_net.py --config-file "configs/pytorch_mask_rcnn_R2_50_s4_FPN_2x.yaml" SOLVER.IMS_PER_BATCH 2 SOLVER.BASE_LR 0.0025 SOLVER.MAX_ITER 720000 SOLVER.STEPS "(480000, 640000)" TEST.IMS_PER_BATCH 1 to train the model as it cannot reproduce the performance of the model zoo in maskrcnn-benchmark.
Actually the maskrcnn-benchmark is deprecated now. We also have the res2net+mmdetection and res2net+detectron2. You can use this two repos to achieve much better performance.
So, now I cannot use single GPU?
So, now I cannot use single GPU?
You cannot reproduce the result with one GPU.
So we cannot train with one GPU you mean?
So we cannot train with one GPU you mean? yes
What should I change if I want to use single GPU? thank you.
Hello Sir, Can you please help me with this problem. I tried many ways but I cannot success in that. Error shows in line 160. Thank you.