Boosting Video Inferencing throughput with Asynchronous Split DNN’s
Budget: $10 – $160 USD
I am doing implementation detecting the video take it from the client to the server as async in Tornado
I am working on a publication paper, and most of what is in the paper has been done. But I need someone to help me modify some codes. I will explain them to you if possible. Just for quick onboarding, what I am doing is splitting a large DL model and running one segment (left) on the client and the other segment (right) on the server.
First, I need to see the total processing time and find out how much time it takes for inputs and outputs. And then I find the time taken; if the frames are divided into different parts, is the time the same or will it be different? Mostly it will differ, and I have implemented it and ready the codes using this Resnet50 model. But I want someone to help me use several models with the help of completing the overleaf writing.
I am working on a publication paper, and most of what is in the paper has been done. But I need someone to help me modify some codes. I will explain them to you if possible. Just for quick onboarding, what I am doing is splitting a large DL model and running one segment (left) on the client and the other segment (right) on the server.
First, I need to see the total processing time and find out how much time it takes for inputs and outputs. And then I find the time taken; if the frames are divided into different parts, is the time the same or will it be different? Mostly it will differ, and I have implemented it and ready the codes using this Resnet50 model. But I want someone to help me use several models with the help of completing the overleaf writing.