Change Weapon Detection Response Object
thelastmattstanding opened this issue · comments
Currently images are sent as base-64 images. Additionally, with a desire to send videos to the front-end for rendering, it may be wise to use a different method to return images and videos for consumption.
Proposal:
- Mount a shared volume between ML and backend server.
- Store predicted results (image or video) from ML model to shared volume and pass file paths to image and video in response.