Inference on a test image

[Andre1998Shuvam]

Hello! Thank you for this work! I would like to know how to run inference of your model on a single image.

[zhihou7]

Hi, Thanks for your interest.

Our method is a two-stage HOI detection approach. Thus, you should first use an additional object detector to obtain the object bounding boxes. Then, you can use our model to infer the interaction between the detected human and objects.

An example can be found in https://github.com/zhihou7/VCL#testing.

Regards,