marcoslucianops
replicating the results in readme.md
I am going through results inside the readme. I wanted to replicate this, can I know how this is done using deepstream.
I got the results from a custom code that I created. It's not public for now.
@marcoslucianops Okay, I am have written code to check the performance using deepstream.
Code I used in Deepstream, to check the FPS.
The below are the results I got, I find that yolov7 is almost equally performing to yolov5. Can let me know the process how would I check if I get greater FPS.
The config files are used as mentioned in github readme, and I have used deepstreamtest1 sample code.
| Name | Model Complexity (No of Layers) | Time(fps) | Accuracy | |
|---|---|---|---|---|
| yolov7 | 640 | 399 | 35 | |
| yolov7x | 640 | 454 | 22 | |
| yolov7-d6 | 1280 | 707 | 20 | |
| yolov7-e6 | 1280 | 619 | 28 | |
| yolov7-w6 | 1280 | 451 | 39 | |
| yolov7-e6e | 1280 | 1037 | 17 | |
| yolov5x | 545 | 19 | ||
| yolov5l | 450 | 33.4 | ||
| yolov5s | 355 | 67 | ||
| yolov5m6 | 461 | 61 |
Okay, I am have written code to check the performance using deepstream.
You can just use the deepstream-app to check the FPS.
I got, I find that yolov7 is almost equally performing to yolov5.
The FPS of the YOLOv7 is very similiar to YOLOv5l.
Can let me know the process how would I check if I get greater FPS.
The FP16 mode can increase the FPS.