I am trying to deploy a YOLOv5 model on AWS lambda for detecting objects in video frames; however, because the video length might be too long, is there any way to chunk the videos and deploy them to Lambda? (I want to do that because of the Lambda runtime limitations). I am assuming that I need to first deploy the model using OpenCv and then deploy the OpenCv on Lambda.
1 Answers
I would create a step before this where you extract the frames from the video using something like ffmpeg. Then you can process one of more frames at a time with Lambda depending on how intensive the model is.
Related