BodyPix 2.0 has been released, with multi-person support and improved accuracy

[SamNoteUser]

BodyPix 2.0 has been released, with multi-person support and improved accuracy (based on ResNet50), a new API, weight quantization, and support for different image sizes.

We are excited to announce the release of BodyPix, an open-source machine learning model which allows for person and body-part segmentation in the browser with TensorFlow.js.

What exactly is person segmentation? In computer vision, image segmentation refers to the technique of grouping pixels in an image into semantic areas typically to locate objects and boundaries. The BodyPix model is trained to do this for a person and twenty-four body parts (parts such as the left hand, front right lower leg, or back torso). In other words, BodyPix can classify the pixels of an image into two categories: 1) pixels that represent a person and 2) pixels that represent background. It can further classify pixels representing a person into any one of twenty-four body parts. This might all make more sense if you try a live demo:-

https://storage.googleapis.com/tfjs-models/demos/body-pix/index.html

[MangoTango]

thanks for sharing 👌🏼

[johnsmith001]

Thank you for sharing the insightful update on BodyPix 2.0! Your informative piece nicely highlights its enhanced features and capabilities. It's impressive to see how companies leverage computer vision and machine learning, like BodyPix, for person and body-part segmentation. This technology's ability to classify pixels for human action recognition  is a significant stride in the field.  As industries evolve, the integration of such advanced models reshapes possibilities, opening avenues for innovative applications in various sectors

[johnsmith001]

Person segmentation in computer vision groups image pixels to identify and differentiate a person from the background, aiding in human activity recognition.