I am new to data science & working on a segmentation model, Basically I need to deploy this segmentation model in android devices using TensorFlow-Lite for real time camera frame segmentation. I used unet model to do that but could not get the accuracy I wanted. After exploring so much I found something about video segmentation but I am bit confuse How video segmentation is different from normal image segmentation? Can somebody explain the differences between these two?
Add a comment
|
2 Answers
$\begingroup$
$\endgroup$
Video segmentation is typically modeled as streaming image segmentation. A video is broken into a series of images, segmentation is applied to each image.
answered Mar 19, 2022 at 23:20
$\begingroup$
$\endgroup$
Brian is right. You need to consider a video as a bunch of concatenated images. 30 FPS (Frame Per Second) video has 30 frames on every second of it. You run your model (segmentation, detection, classification or any other vision models does not matter) on the frames within the video stream.
UNet is a heavy model to deploy on a mobile so you've tried a pruned version of it. The lack of accuracy might be caused of the model you used, the angle of camera. You might want control your use case. If low accuracy still remain you try other segmentation models like Yolov8 (might have pruned versions to deploy on edge & mobile), Faster-RCNN etc.
Cheers
