The scenario
A film restoration studio needs to colorize a large batch of black-and-white films using machine learning. The workflow involves uploading films, splitting into frames, processing each frame through a colorization model, reassembling, and final review.
What is in this drawing
Read the decisions behind it.
01
Use of asynchronous batch processing to handle large volumes
02
Integration of ML model as a separate service for scalability
03
Storing intermediate frames in a database for fault tolerance
This sequence applies to any batch image/video processing pipeline where individual frames are transformed by a service and reassembled, especially when using machine learning models.
More like this