Kappa coefficients for different algorithms. حسب Yingying Liu (360782)

"…The actual accuracy and mean latency time of the model were 92.43% and 260ms, respectively. …"

The structure of ASPP+ block. حسب Yingying Liu (360782)

"…The actual accuracy and mean latency time of the model were 92.43% and 260ms, respectively. …"

The structure of attention gate block [31]. حسب Yingying Liu (360782)

"…The actual accuracy and mean latency time of the model were 92.43% and 260ms, respectively. …"

DSC block and its application network structure. حسب Yingying Liu (360782)

"…The actual accuracy and mean latency time of the model were 92.43% and 260ms, respectively. …"

The structure of multi-scale residual block [30]. حسب Yingying Liu (360782)

"…The actual accuracy and mean latency time of the model were 92.43% and 260ms, respectively. …"

The structure of IRAU and Res2Net+ block [22]. حسب Yingying Liu (360782)

"…The actual accuracy and mean latency time of the model were 92.43% and 260ms, respectively. …"

Dataset visualization diagram. حسب Yaojun Zhang (389482)

"…Results on a specialized dataset reveal that YOLOv8n-BWG outperforms YOLOv8n by increasing the mean Average Precision (mAP) by 4.2%, boosting recognition speed by 21.3% per second, and decreasing both the number of floating-point operations (FLOPs) by 28.9% and model size by 26.3%. …"

Dataset sample images. حسب Yaojun Zhang (389482)

"…Results on a specialized dataset reveal that YOLOv8n-BWG outperforms YOLOv8n by increasing the mean Average Precision (mAP) by 4.2%, boosting recognition speed by 21.3% per second, and decreasing both the number of floating-point operations (FLOPs) by 28.9% and model size by 26.3%. …"

Performance comparison of different models. حسب Yaojun Zhang (389482)

"…Results on a specialized dataset reveal that YOLOv8n-BWG outperforms YOLOv8n by increasing the mean Average Precision (mAP) by 4.2%, boosting recognition speed by 21.3% per second, and decreasing both the number of floating-point operations (FLOPs) by 28.9% and model size by 26.3%. …"

C2f and BC2f module structure diagrams. حسب Yaojun Zhang (389482)

"…Results on a specialized dataset reveal that YOLOv8n-BWG outperforms YOLOv8n by increasing the mean Average Precision (mAP) by 4.2%, boosting recognition speed by 21.3% per second, and decreasing both the number of floating-point operations (FLOPs) by 28.9% and model size by 26.3%. …"

YOLOv8n detection results diagram. حسب Yaojun Zhang (389482)

"…Results on a specialized dataset reveal that YOLOv8n-BWG outperforms YOLOv8n by increasing the mean Average Precision (mAP) by 4.2%, boosting recognition speed by 21.3% per second, and decreasing both the number of floating-point operations (FLOPs) by 28.9% and model size by 26.3%. …"

YOLOv8n-BWG model structure diagram. حسب Yaojun Zhang (389482)

"…Results on a specialized dataset reveal that YOLOv8n-BWG outperforms YOLOv8n by increasing the mean Average Precision (mAP) by 4.2%, boosting recognition speed by 21.3% per second, and decreasing both the number of floating-point operations (FLOPs) by 28.9% and model size by 26.3%. …"

BiFormer structure diagram. حسب Yaojun Zhang (389482)

"…Results on a specialized dataset reveal that YOLOv8n-BWG outperforms YOLOv8n by increasing the mean Average Precision (mAP) by 4.2%, boosting recognition speed by 21.3% per second, and decreasing both the number of floating-point operations (FLOPs) by 28.9% and model size by 26.3%. …"

YOLOv8n-BWG detection results diagram. حسب Yaojun Zhang (389482)

"…Results on a specialized dataset reveal that YOLOv8n-BWG outperforms YOLOv8n by increasing the mean Average Precision (mAP) by 4.2%, boosting recognition speed by 21.3% per second, and decreasing both the number of floating-point operations (FLOPs) by 28.9% and model size by 26.3%. …"

GSConv module structure diagram. حسب Yaojun Zhang (389482)

"…Results on a specialized dataset reveal that YOLOv8n-BWG outperforms YOLOv8n by increasing the mean Average Precision (mAP) by 4.2%, boosting recognition speed by 21.3% per second, and decreasing both the number of floating-point operations (FLOPs) by 28.9% and model size by 26.3%. …"

Performance comparison of three loss functions. حسب Yaojun Zhang (389482)

"…Results on a specialized dataset reveal that YOLOv8n-BWG outperforms YOLOv8n by increasing the mean Average Precision (mAP) by 4.2%, boosting recognition speed by 21.3% per second, and decreasing both the number of floating-point operations (FLOPs) by 28.9% and model size by 26.3%. …"

mAP0.5 Curves of various models. حسب Yaojun Zhang (389482)

"…Results on a specialized dataset reveal that YOLOv8n-BWG outperforms YOLOv8n by increasing the mean Average Precision (mAP) by 4.2%, boosting recognition speed by 21.3% per second, and decreasing both the number of floating-point operations (FLOPs) by 28.9% and model size by 26.3%. …"

Network loss function change diagram. حسب Yaojun Zhang (389482)

"…Results on a specialized dataset reveal that YOLOv8n-BWG outperforms YOLOv8n by increasing the mean Average Precision (mAP) by 4.2%, boosting recognition speed by 21.3% per second, and decreasing both the number of floating-point operations (FLOPs) by 28.9% and model size by 26.3%. …"

Comparative diagrams of different indicators. حسب Yaojun Zhang (389482)

"…Results on a specialized dataset reveal that YOLOv8n-BWG outperforms YOLOv8n by increasing the mean Average Precision (mAP) by 4.2%, boosting recognition speed by 21.3% per second, and decreasing both the number of floating-point operations (FLOPs) by 28.9% and model size by 26.3%. …"

YOLOv8n structure diagram. حسب Yaojun Zhang (389482)

"…Results on a specialized dataset reveal that YOLOv8n-BWG outperforms YOLOv8n by increasing the mean Average Precision (mAP) by 4.2%, boosting recognition speed by 21.3% per second, and decreasing both the number of floating-point operations (FLOPs) by 28.9% and model size by 26.3%. …"