YOLOv9 vs YOLOv8 Performance Comparison on Fire and Smoke Detection task (Day scenarios)

🔥🔥🔥NEW Fire and Smoke Dataset for Object Detection🔥🔥🔥

I utilized this be dataset to train several YOLO models, including YoloV5, YoloV6, YoloV7, YoloV8, YoloV9 and YoloNAS.

All my training, inference, and evaluation results, along with the dataset, are publicly accessible on GitHub and GoogleDrive.

I also created a script that generate a detailed metric after each inference task to evaluate the performance of each model by examining the number of detections and the average confidence for each class. In this way is easier to determine which model is the best for this task of fire and smoke detection.

👉 Roboflow Datasets:

🤖 GitHub repo:

👉 LinkedIn:

👉 Google Drive:

#YOLO hashtag#Roboflow hashtag#fire hashtag#smoke #DeepLearnjng #Dataset #train #inference #objectdetection #github

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

Evaluation Metric:

yolov5m yolov6m yolov7 yolov8m yolov9-c yv9gelan-c
--------------------------------------------------------------------------------------------------------------------------------------------
All Detections 35980 39119 38585 34851 33847 37707
Avg FPS 207 214 179 197 115 98
Avg InfTime 5.60 ms 4.98 ms 6.35 ms 5.30 ms 8.88 ms 10.38 ms
Conf_thr 0.4 0.4 0.4 0.4 0.4 0.4
Total InfTime 1046.08 s 930.26 s 1186.18 s 990.04 s 1658.78 s 1938.98 s
--------------------------------------------------------------------------------------------------------------------------------------------