Segmentation Fault while running demo

[gsan2]

Hi

I am facing the following issue when I try to run the demo on a Ubuntu machine with GPU on it. (FairMOT) @ubuntu:~/ML/FairMOT/src$ python demo.py mot --load_model /home/ML/FairMOT/models/all_dla34.pth --conf_thres 0.4

Fix size testing. training chunk_sizes: [6, 6] The output will be saved to /home/ML/FairMOT/src/lib/../../exp/mot/default heads {'hm': 1, 'wh': 2, 'id': 512, 'reg': 2} 2020-04-16 10:58:12 [INFO]: Starting tracking... Lenth of the video: 1500 frames Creating model... loaded /home/ML/FairMOT/models/all_dla34.pth, epoch 10 2020-04-16 10:58:16 [INFO]: Processing frame 0 (100000.00 fps) Fatal Python error: Segmentation fault

Current thread 0x00007f53c3767740 (most recent call first): return self File "/home/ML/FairMOT/src/lib/models/networks/DCNv2/dcn_v2.py", line 31 in forward File "/home/ML/FairMOT/src/lib/models/networks/DCNv2/dcn_v2.py", line 128 in forward File "/home/anaconda3/envs/FairMOT/lib/python3.7/site-packages/torch/nn/modules/module.py", line 547 in call File "/home/ML/FairMOT/src/lib/models/networks/pose_dla_dcn.py", line 355 in forward File "/home/anaconda3/envs/FairMOT/lib/python3.7/site-packages/torch/nn/modules/module.py", line 547 in call File "/home/ML/FairMOT/src/lib/models/networks/pose_dla_dcn.py", line 384 in forward File "/home/anaconda3/envs/FairMOT/lib/python3.7/site-packages/torch/nn/modules/module.py", line 547 in call File "/home/ML/FairMOT/src/lib/models/networks/pose_dla_dcn.py", line 411 in forward File "/home/anaconda3/envs/FairMOT/lib/python3.7/site-packages/torch/nn/modules/module.py", line 547 in call File "/home/ML/FairMOT/src/lib/models/networks/pose_dla_dcn.py", line 472 in forward File "/home/anaconda3/envs/FairMOT/lib/python3.7/site-packages/torch/nn/modules/module.py", line 547 in call File "/home/ML/FairMOT/src/lib/tracker/multitracker.py", line 246 in update File "/home/ML/FairMOT/src/track.py", line 63 in eval_seq File "demo.py", line 33 in demo self.height = img_size[1] File "demo.py", line 45 in Segmentation fault (core dumped)

Could you please help with this? Thanks! GK

[jamessmith123456]

i got it too..Segmentation fault(core dumped). Have you solve it?

[ifzhang]

Have you successfully built DCNv2？I think the error is caused by "You're using a third-party extension module written in C, and that extension module has crashed." You can try to build DCNv2 again or use our HRNetW18 model which does not need DCNv2 to run the demo (don't forget to comment the lines with 'dcn' in src/lib/models/model.py).

[toyal]

i got it too..Segmentation fault(core dumped). Have you solve it?

have you solve this problem,i locate the problem caused by dcnv2,but how to solve it

[ZhangYuef]

Have you successfully built DCNv2？I think the error is caused by "You're using a third-party extension module written in C, and that extension module has crashed." You can try to build DCNv2 again or use our HRNetW18 model which does not need DCNv2 to run the demo (don't forget to comment the lines with 'dcn' in src/lib/models/model.py).

@ifzhang How to use HRNetW18 model correctly? I use following command:

python track.py mot --load_model ../models/hrnetv2_w18_imagenet_pretrained.pth --conf_thres 0.6 --val_mot16 False

Got following warning:

No param reg.2.weight.If you see this, your model does not fully load the pre-trained weight. Please make sure you have correctly specified --arch xxx or set the correct --num_classes for your own dataset. No param reg.2.bias.If you see this, your model does not fully load the pre-trained weight. Please make sure you have correctly specified --arch xxx or set the correct --num_classes for your own dataset.

and all NaN value ...