AlexeyAB / Yolo_mark

GUI for marking bounded boxes of objects in images for training neural network Yolo v3 and v2
https://github.com/AlexeyAB/darknet
The Unlicense
1.8k stars 680 forks source link

How to increase the FPS in real-time streaming? #211

Open PIjarihd opened 2 years ago

PIjarihd commented 2 years ago

I am using YOLOv4 for custom object detection. I have 68 classes and the number of trining images is more than 16k, and the size of weight file is 257 mb. I used YOLOv4 on real time streaming on CCTV using Jetson tx2 board. The frmae rate is from 3.1 to 4.9 fps. When using YOLOv5s, the frame rate is more than 20 fps. I need to increase the fps and I want to use yolov4 version. What should I do?

Should I use YOLOv4tiny?