Pytorch version. Visualize the results

[jgallegov]

Hello, I am using the pythorch version, and I can't see any option to visualize the results. Please, can you tell me how to proceed with demo_fast.py?

Thank you

[nerd-qp]

I only find a vis_image.py in the code, which visualize the bounding box. Though, I have to do some hacking for it to work.

[jgallegov]

Hello, then there is not any args to allow visualization¿?

[Fang-Haoshu]

Hi, we will add that soon

[Jeff-sjtu]

You can visualize the result now!

python demo_fast.py \
--inputlist ./list-coco-minival500.txt \
--imgpath ${img_directory} \
--outputpath ./coco-minival \
--vis_res True

The visualization results will be saved in outputpath.

[huangzhenjie]

hello， I can visualize the result now, it look great!! and I want to know whether pytorch-version will support video in the future? or it whether support Pose Tracking? tkanks, have a good day.

[jgallegov]

Hello,

I've got this error:

Loading SSD model.. Traceback (most recent call last): File "demo_fast.py", line 33, in det_model.load_state_dict(torch.load('./models/ssd/ssd_coco.pth')) File "C:\Anaconda3\envs\tensorflow\lib\site-packages\torch\nn\modules\module.py", line 706, in load_state_dict self.class.name, "\n\t".join(error_msgs))) RuntimeError: Error(s) in loading state_dict for FPNSSD512: Missing key(s) in state_dict: "fpn.bn1.num_batches_tracked", "fpn.layer1.0.bn1.num_batches_tracked", "fpn.layer1.0.bn2.num_batches_tracked", "fpn.layer1.0.bn3.num_batches_tracked", "fpn.layer1.0.downsample.1.num_batches_tracked", "fpn.layer1.1.bn1.num_batches_tracked", "fpn.layer1.1.bn2.num_batches_tracked", "fpn.layer1.1.bn3.num_batches_tracked", "fpn.layer1.2.bn1.num_batches_tracked", "fpn.layer1.2.bn2.num_batches_tracked", "fpn.layer1.2.bn3.num_batches_tracked", "fpn.layer2.0.bn1.num_batches_tracked", "fpn.layer2.0.bn2.num_batches_tracked", "fpn.layer2.0.bn3.num_batches_tracked", "fpn.layer2.0.downsample.1.num_batches_tracked", "fpn.layer2.1.bn1.num_batches_tracked", "fpn.layer2.1.bn2.num_batches_tracked", "fpn.layer2.1.bn3.num_batches_tracked", "fpn.layer2.2.bn1.num_batches_tracked", "fpn.layer2.2.bn2.num_batches_tracked", "fpn.layer2.2.bn3.num_batches_tracked", "fpn.layer2.3.bn1.num_batches_tracked", "fpn.layer2.3.bn2.num_batches_tracked", "fpn.layer2.3.bn3.num_batches_tracked", "fpn.layer3.0.bn1.num_batches_tracked", "fpn.layer3.0.bn2.num_batches_tracked", "fpn.layer3.0.bn3.num_batches_tracked", "fpn.layer3.0.downsample.1.num_batches_tracked", "fpn.layer3.1.bn1.num_batches_tracked", "fpn.layer3.1.bn2.num_batches_tracked", "fpn.layer3.1.bn3.num_batches_tracked", "fpn.layer3.2.bn1.num_batches_tracked", "fpn.layer3.2.bn2.num_batches_tracked", "fpn.layer3.2.bn3.num_batches_tracked", "fpn.layer3.3.bn1.num_batches_tracked", "fpn.layer3.3.bn2.num_batches_tracked", "fpn.layer3.3.bn3.num_batches_tracked", "fpn.layer3.4.bn1.num_batches_tracked", "fpn.layer3.4.bn2.num_batches_tracked", "fpn.layer3.4.bn3.num_batches_tracked", "fpn.layer3.5.bn1.num_batches_tracked", "fpn.layer3.5.bn2.num_batches_tracked", "fpn.layer3.5.bn3.num_batches_tracked", "fpn.layer4.0.bn1.num_batches_tracked", "fpn.layer4.0.bn2.num_batches_tracked", "fpn.layer4.0.bn3.num_batches_tracked", "fpn.layer4.0.downsample.1.num_batches_tracked", "fpn.layer4.1.bn1.num_batches_tracked", "fpn.layer4.1.bn2.num_batches_tracked", "fpn.layer4.1.bn3.num_batches_tracked", "fpn.layer4.2.bn1.num_batches_tracked", "fpn.layer4.2.bn2.num_batches_tracked", "fpn.layer4.2.bn3.num_batches_tracked".

Any idea about?

Thanks

[Jeff-sjtu]

Hi, @huangzhenjie. We will consider video-input or camera-input in future.

[Jeff-sjtu]

@jgallegov Did you use the latest code?

[Hansoluo]

@jgallegov it seems like a pytorch version problem. you can see more info in Unexpected key(s) in state_dict. I think you can update the PyTorch to newer version.