The current API is unique and forks to different processes depending on the payload (file) type.
It should be changed to different routes:
- Video route => outputs a json with the detected trashes
- Zip route => same as video
- Image route => demo route, outputs a new image with the detected objects bounded by boxes drawn upon it