Hi, I am running an application where I need to generate keypoints of humans from real-time feed of webcams. I am using yolov7.pt for detecting human bboxes and yolov7-w6-pose.pt for generating keypoints of humans. I have to run two of these models in parallel threads because when using just the yolov7-w6-pose.pt model for generating keypoints as well as bboxes, I get ghost keypoints, i.e. keypoints on blank spaces, such as chairs, tables etc. What might be the reason behind this.
Hi, I am running an application where I need to generate keypoints of humans from real-time feed of webcams. I am using yolov7.pt for detecting human bboxes and yolov7-w6-pose.pt for generating keypoints of humans. I have to run two of these models in parallel threads because when using just the yolov7-w6-pose.pt model for generating keypoints as well as bboxes, I get ghost keypoints, i.e. keypoints on blank spaces, such as chairs, tables etc. What might be the reason behind this.