YoloV9 vs YoloV8 vs YoloV7 Inference Performance Comparison on Real-World Scenarios

In this research, I analyze the performance of YoloV7, YoloV8, and YoloV9 in terms of inference speed and accuracy. The analysis uses models weights that were pretrained on the COCO dataset

For the performance comparison, I use specific metrics generated after each inference task. These metrics include the total number of detections for each class and the average confidence for individual classes. This approach helps determine which model is most effective at detecting specific instances in images. For more details about these scripts, please refer to the GitHub repository mentioned bellow.


The results of the experiment are stored in GoogleDrive at path:

#yolo #yolov7 #yolov8 #yolov9 #objectdetection
#deeplearning #ai #inference #ai #computervision #tutorials #rtx4090

Thank you for watching! 😄
I hope you enjoyed the experiment!

👉 LinkedIn :

Evaluation Metric:

yolov7 yolov8l yolov9-c yv9gelan-c
--------------------------------------------------------------------------------------------------------
All Detections 68 399 56 224 56 395 100 036
Avg FPS 317 113 87 75
Avg InfTime 3.16 ms 9.51 ms 12.12 ms 13.77 ms
Conf_thr 0.4 0.4 0.4 0.4
Total InfTime 106.40 s 320.20 s 408.08 s 463.64 s
-----------------------------------------------------------------------------------------------------------