Computer Vision News - May 2022

11 MediaPipe 2/2 A peak into the integration with Javascript Let’s finish this articlewith a brief overviewof theMediaPipe solutions availableon Javascript and how to access them. Youwill find allML pipelines availablewithMediaPipe staged inNPM. Hence, if youdownload this from here , you can just proceed by installing any package locally using npm install . npm install @mediapipe/face _ mesh npm install @mediapipe/drawing _ utils npm install @mediapipe/camera _ utils npm install @mediapipe/control _ utils The commands above install the face mesh solution, the utilities to draw landmarks and connectors, to operate the camera and to show sliders and FPS widgets. Another great service offered by MediaPipe are the demos on Codepen. You can find them here and test any solution live. Through the button on the top right, you can alternate between a full-page view for testing the pipelines or an editor view which shows you the HTML, CSS and JS codes. On the left of the Codepen window, you can find sliders to change parameters of the MediaPipe app. We introduced these for the Hands demo in the former article. The max_num_hands can be used to set the maximum number of hands to detect, model_complexity , min_detection_confidence and min_tracking_confidence to vary the lower bound confidence for both the hand and landmark detection. Figure 3: Full-page view of the Hands demo on Codepen.