Cannot test with the pre-trained Detectron model?

[baiyancheng20]

I use the following commands for inferring: python tools/infer_simple.py --dataset coco --cfg ./configs/baselines/e2e_faster_rcnn_R-50-C4_2x.yaml --load_detectron ./model_final.pkl --image_dir ./demo/sample_images --output_dir ./demo/

I got the following error: p_tensor.copy_(torch.Tensor(src_blobs[d_name])) KeyError: 'conv_rpn_w'

I checked the src_blobs in detectron_weight_helper.py and I found the key and value of 'conv_rpn_w' didn't exit. Could you help me deal with this?

• Operating system: Ubuntu 16.04
• CUDA version: 8
• cuDNN version: 7
• GPU models (for all devices if they are not all the same): 1080t9
• python version: 3.6
• pytorch version: 0.4
• Anything else that seems relevant: ?

[baiyancheng20]

Problem has been solved. Download the wrong model.

[XPandora]

Hi, I have the same problem, can you give a correct model link?

[MindyZLM]

@XPandora Hi, I have the same problem,can you tell me how to solve it? thanks!