vov-net / vovnet-fcos Goto Github PK

Hi @stigma0617 .

I noticed that this implementation is still utilizing the code from maskrcnn_benchmark but you created a fcos core version of maskrcnn_benchmark that leave only the fcos related component.

cd ../VoVNet-FCOS/
python setup.py build develop
....
Finished processing dependencies for fcos==0.1.9

However, after I rebuilt successfully with the above commands. When I tired to train with my own dataset. I encountered Cpython error.

ImportError: ../VoVNet-FCOS/fcos_core/_C.cpython-37m-x86_64-linux-gnu.so: undefined symbol: _ZN3c105ErrorC1ENS_14SourceLocationERKSs

Can you advise?

hi@stigma0617
when i train the code get error like this,can you help me?thanks

File "tools/train_net.py", line 180, in
main()
File "tools/train_net.py", line 173, in main
model = train(cfg, args.local_rank, args.distributed)
File "tools/train_net.py", line 31, in train
model = build_detection_model(cfg)
File "/home/yewei/wbq/FCOS/fcos_core/modeling/detector/detectors.py", line 10, in build_detection_model
return meta_arch(cfg)
File "/home/yewei/wbq/FCOS/fcos_core/modeling/detector/generalized_rcnn.py", line 29, in init
self.backbone = build_backbone(cfg)
File "/home/yewei/wbq/FCOS/fcos_core/modeling/backbone/backbone.py", line 101, in build_backbone
cfg.MODEL.BACKBONE.CONV_BODY
AssertionError: cfg.MODEL.BACKBONE.CONV_BODY: V-39-FPN-RETINANET are not registered in registry
Traceback (most recent call last):
File "/home/yewei/anaconda3/envs/fcos/lib/python3.7/runpy.py", line 193, in _run_module_as_main
"main", mod_spec)
File "/home/yewei/anaconda3/envs/fcos/lib/python3.7/runpy.py", line 85, in _run_code
exec(code, run_globals)
File "/home/yewei/anaconda3/envs/fcos/lib/python3.7/site-packages/torch/distributed/launch.py", line 246, in
main()
File "/home/yewei/anaconda3/envs/fcos/lib/python3.7/site-packages/torch/distributed/launch.py", line 242, in main
cmd=cmd)
subprocess.CalledProcessError: Command '['/home/yewei/anaconda3/envs/fcos/bin/python', '-u', 'tools/train_net.py', '--local_rank=0', '--config-file', 'configs/vovnet/fcos_V_39_FPN_1x.yaml']' returned non-zero exit status 1.

I cann't get paraments of the Fcos_V_39xxx from China. Can you show me how to train from random initialization?

@stigma0617
how to train keypoint by FCOS, I get error use this

BACKBONE:
CONV_BODY: "V-39-eSE-FPN-RETINANET"
FREEZE_CONV_BODY_AT: 0
VOVNET:
BACKBONE_OUT_CHANNELS: 128
RPN_ONLY: False
FCOS_ON: True
KEYPOINT_ON: True
RETINANET:
USE_C5: False # FCOS uses P5 instead of C5
ROI_HEADS:
USE_FPN: True
ROI_KEYPOINT_HEAD:
SHARE_BOX_FEATURE_EXTRACTOR: True