HandTrack.js
Hand Palm Detector
RUN
This module detects key points in hand figures using Handtrack.js. The model works by predicting bounding boxes around each hand in a frame of a video recording. It then outputs a confidence score and the coordinates of each bounding box. The confidence score represents the probability that the hand has been correctly detected. It ranges from a value of 0 to 1, where 1 represents an exact detection. The bounding box is represented by the top-left corner coordinates of the box, followed by the width and height of the box.
The data is exported as a CSV file structured as follows:
| frame | hand | confidence | x | y | width | height |
|---|---|---|---|---|---|---|
| 0 | 0 | 0.769430101 | 134.9761105 | 257.2151423 | 150.4400349 | 225.1269221 |
| 0 | 1 | 0.7788378 | 138.6055374 | 256.8473518 | 148.7379074 | 226.7438173 |
| 1 |
The frame column represents the frame number in the video. Each frame consists of a number of hands represented in the hand column. Each hand is represented with a confidence score and key points of its bounding box. For example, person 1 in frame 0 has a confidence score of 0.78, its bounding box’s top-left coordinate is (138.60, 256.85), and the width and height of the bounding box is 148.74 and 226.74 respectively. Some frames will not contain any values. This means that no hand was detected in the frame. So, in frame 1 of the table, no hand was detected by the model.