search

mks0601 / 3DMPPE_ROOTNET_RELEASE

Official PyTorch implementation of "Camera Distance-aware Top-down Approach for 3D Multi-person Pose Estimation from a Single RGB Image", ICCV 2019
MIT License
477 stars 65 forks source link

MuPots Evaluation #25

Open nicolasugrinovic opened 3 years ago

nicolasugrinovic commented 3 years ago

Hi, I have a question regarding MuPots annotations. I know you did not create the dataset. However, I see that annotations for some people in the images are missing. For example, for TS1/img_0000001.jpg, the image has 3 people, but annotations for 2d and 3d_kpts come only for 2 people.

Have you came with this problem while evaluating your method?

Thanks

mks0601 commented 3 years ago

I checked that the evaluation is only performed for the annotated persons. As the evaluation metric is 3DPCK, which does not consider false positive detections, I think it would not matter.

mks0601 commented 3 years ago

I cannot remember clearly, but I think you'd better use the evaluation script. You can download it from the official dataset page. The script outputs several .csv files, which contain the evaluation results including 3DPCKs. I used the .csv files to report the performance of my method.

nicolasugrinovic commented 3 years ago

Ok thank you