Day 81 – COCO Keypoint Detection Detectron2 Computer Vision by Facebook AI Research (FAIR)

We have seen the below posts on Detectron2.

• Detectron2, Algorithms, Features, and code on Mask R-CNN
• COCO Panoptic Segmentation Detectron2 Computer Vision by Facebook AI Research (FAIR)

Today, we will see in this blog about COCO Keypoint Detection in Detectron2.

What is COCO Keypoint Detection?

The COCO Keypoint Detection Task requires localization of person keypoints in challenging, uncontrolled conditions. The keypoint task involves simultaneously detecting people and localizing their keypoints (person locations are not given at test time)

Source: COCO Keypoint Detection

Image Source: COCO Keypoint Detection

The above image gives you a clear view of Keypoint detection.

COCO dataset contains more than 200,000 images and 250,000 persons labeled with keypoints.  Annotations on train and validation are available for public and it contains more than 150,000 persons and 1,700,000  labeled keypoints.

Let’s copy existing code from Detectron2 and test the COCO Keypoint Detection.

I am using Google Colab and you check the online code here.

Output:

Detectron2 COCO Keypoint Detection labels only persons unlike Panoptic FPN or Mask R-CNN.

What is your thought on this Detectron2 COCO Keypoint Detection?

Would you please comment below and where we can use in future technologies?

Links:

Google Colab COCO Keypoint Detection

Detectron2 – Facebook AI

Detectron2 Github Repository