This container makes it possible to quickly deploy a pretrained, highly-efficient Computer Vision model. You can send any format of image or video - PNG, JPG, MP4 and AVI will be processed directly, the container will automatically convert anything else to one of those. Metadata is outputted as JSON. This container can use any of the sizes of YoloX - namely X, L, M, S, Tiny and Nano. This container returns bounding-boxed results for each frame in JSON format.

There is a doc link with instructions to use the product in the Support description section