VidHOI
VidHOI copied to clipboard
coldmanck
Official implementation of "ST-HOI: A Spatial-Temporal Baseline for Human-Object Interaction Detection in Videos" (ACM ICMRW 2021)
Hello, the Baidu web disk link of data set training videos (part 1-4) is invalid, can you update it
 in file https://github.com/coldmanck/VidHOI/blob/master/vidor_eval.ipynb. We should not just skip these triplets. Instead, ap for these triplets should be set as 0.
Hello, @coldmanck Thank you for providing your work. I trained the model with your guide. by the way, Can you tell me how to train the model with smaller action...
Errors occur during training and Validation：'Tensor' object has no attribute 'tensor'
Niether "demo_net_vidor.py" nor "demo_net.py" do not seem to be using Skeleton based information for visualization. Could you please provide some details on demo?
Hi, I would like to know how to get the HOI categories list, which include 557 triplets. Thanks!
Hi. I have a question when I using the vidor_eval.ipynb script to generate mAP. The script seems to only support single-label case? If a ground-truth human-object pair has multiple interactions,...
