WARNING [01/11 17:21:52 d2.config.compat]: Config '/home/lvxinyu/AVVP/Others/py-bottom-up-attention/configs/VG-Detection/faster_rcnn_R_101_C4_caffe.yaml' has no VERSION. Assuming it to be compatible with latest v2. Modifications for VG in RPN (modeling/proposal_generator/rpn.py): Use hidden dim 512 instead fo the same dim as Res4 (1024). Modifications for VG in RoI heads (modeling/roi_heads/roi_heads.py): 1. Change the stride of conv1 and shortcut in Res5.Block1 from 2 to 1. 2. Modifying all conv2 with (padding: 1 --> 2) and (dilation: 1 --> 2). For more details, please check 'https://github.com/peteanderson80/bottom-up-attention/blob/master/models/vg/ResNet-101/faster_rcnn_end2end_final/test.prototxt'.

Question

WARNING [01/11 17:21:52 d2.config.compat]: Config '/home/lvxinyu/AVVP/Others/py-bottom-up-attention/configs/VG-Detection/faster_rcnn_R_101_C4_caffe.yaml' has no VERSION. Assuming it to be compatible with latest v2. Modifications for VG in RPN (modeling/proposal_generator/rpn.py): Use hidden dim 512 instead fo the same dim as Res4 (1024). Modifications for VG in RoI heads (modeling/roi_heads/roi_heads.py): 1. Change the stride of conv1 and shortcut in Res5.Block1 from 2 to 1. 2. Modifying all conv2 with (padding: 1 --> 2) and (dilation: 1 --> 2). For more details, please check 'https://github.com/peteanderson80/bottom-up-attention/blob/master/models/vg/ResNet-101/faster_rcnn_end2end_final/test.prototxt'.

XinyuLyu opened this issue 4 years ago · 3 comments

Hi! When I run the demo_vg_detection.ipynb, it has such logs without predicted results. Can you help me with that?

WARNING [01/11 17:21:52 d2.config.compat]: Config '/home/lvxinyu/AVVP/Others/py-bottom-up-attention/configs/VG-Detection/faster_rcnn_R_101_C4_caffe.yaml' has no VERSION. Assuming it to be compatible with latest v2.
Modifications for VG in RPN (modeling/proposal_generator/rpn.py):
Use hidden dim 512 instead fo the same dim as Res4 (1024).

Modifications for VG in RoI heads (modeling/roi_heads/roi_heads.py):
1. Change the stride of conv1 and shortcut in Res5.Block1 from 2 to 1.
2. Modifying all conv2 with (padding: 1 --> 2) and (dilation: 1 --> 2).
For more details, please check 'https://github.com/peteanderson80/bottom-up-attention/blob/master/models/vg/ResNet-101/faster_rcnn_end2end_final/test.prototxt'.

Answer 1 · 2021-01-11T18:10:22.000Z

Thanks.

I think that the reason might be a slow speed in downloading the weights. Could you help test whether you could download the weight file http://nlp.cs.unc.edu/models/faster_rcnn_from_caffe.pkl locally? If so, you could edit the line

cfg.MODEL.WEIGHTS = "http://nlp.cs.unc.edu/models/faster_rcnn_from_caffe.pkl"

with this local downloaded location and it should work fine now.

Best,
Hao

Answer 2 · 2021-01-12T08:43:37.000Z

Thank you for your reply. I have tried it before, but still not work. Actually I modify '.ipynb' to '.py', here is my code and the results, does it matters?

Bu the way, when I debug, it crash here with no logs.

Answer 3 · 2021-01-12T18:08:08.000Z

I do not think that these modifications to py would cause a problem.

Btw, the current return message is a WARNING which says that the version of configuration is not specified. This WARNING would not cause an issue in other processing steps.
Have you encountered any other issues when you proceed with this script?

I encourage you to test with the original ipynb once to see whether the outputs are successfully printed. It should work without CUDA as well.