search

niklongstone / rekognition-video-tracking

Uses Amazon Rekognition to get bounding boxes around detected object.
1 stars 0 forks source link

play.py doesn't display any bounding box from the json file. #1

Open progerjkd opened 3 years ago

progerjkd commented 3 years ago

Hi there,

interesting project. I'm able to use detect_video.py to generate the JSON file containing the labels of detected objects and scenes. But when using play.py I don't see any bounding box being drawn on the output video.

$ python3 detect_video.py -v NEW2_small.mp4 -b rekognition-data $ python3 play.py -v NEW2_small.mp4 -d labels/NEW2_small.mp4.json -f 10

Any advice?

thanks in advance. Roger

niklongstone commented 3 years ago

Hi Roger,

to summarise in short how it works: the library parses the GetLabelDetection response looking for a label called Person. If present, bounding boxes are extracted from Instances -> BoundingBox and displayed. It might be possible that the json doesn't contain Person or is malformed.