RecoMadeEasy ® Object Detection Applied on a Soccer Hightlights video
This demo was run at 30 frames per second detection rate on a Lenovo laptop all locally with no aid of any external computation aid. It uses the RecoMadeEasy ® Large Object Detection model. The object detection is done on every frame of the original video in realtime. The output is in the form of a JSON file which is shown at the bottom of the clip. The top match for all objects with a posterior probability higher than 30%, including the predicted size and location of the objects, are shown in boxes with the label printed on top.