Jincan Zhu

Jincan Zhu

Auto-generated profile, based on publicly available information

Jincan Zhu

College of Big Data and Intelligent Engineering, Southwest Forestry University, Kunming, China

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
7 October 2024 in IEEE Access
Verified
Journal Article
Open access
JZJincan Zhu
Jincan Zhu
CMChenhao Ma
Chenhao Ma
JRJian Rong
Jian Rong
YCYong Cao
Yong Cao
At present, the protection of birds, especially endangered birds, faces major challenges. In the process of protection, birds are often mixed with various drones, and it is difficult to accurately count the number of endangered birds, which brings great difficulties to bird protection work. So tracking and identifying birds and drones is crucial. To solve these problems, this paper proposes a new multi-target tracking (MOT) model based on the combination of YOLOv9 detection algorithm and DeepSORT tracking algorithm. Firstly, the original RepNSCPELAN4 module is replaced by CAM context feature enhancement module in Backbone to improve the model’s ability to extract small target features. Following this, the AFF channel attention mechanism has been integrated with RepNSCPELAN4 in the Head section to create the Repnscpelan4-AFF module, which aims to better address semantic and scale inconsistencies. Finally, a new RepNSCPELAN4-Akconv module has been developed using the AKConv dynamic Convolution module to replace the RepNSCPELAN4 module in the original Head section, enabling the model to more effectively capture detailed and contextual information. On the Bird-Drone visible light comprehensive data set proposed in this study, the improved YOLOv9-DeepSORT model has a mAP0.50 of 81.3% for all categories and 89.1% for individual birds. Compared to the baseline YOLOv9 original model, improvements of 7.9% and 23.9% respectively. On infrared datasets, compared to the original model, the mAP0.50 of the improved model is improved by 3.2% in all categories. The accuracy of identifying individual birds and similarly shaped fixed-wing drones also improved by 2.2% and 7.5% respectively. Moreover, on the mixed visible light and infrared data sets, the model get mAP0.50 of 81.8% higher 0.9% than that of the YOLOv9. These experiments demonstrate the improved YOLOv9-DeepSORT method can expand the multiscene application range of bird recognition and tracking models, effectively promoting the extraction of video frame features in multi-target tracking.

15 April 2025 in IEEE Access
Verified
Journal Article
Open access
PSPeichun Suo
Peichun Suo
JZJincan Zhu
Jincan Zhu
QZQingyang Zhou
Qingyang Zhou
Weili KouWeili Kou
Weili Kou
XWXiuyan Wang
Xiuyan Wang
WSWen Suo
Wen Suo
The ability to distinguish between birds and drones is essential for applications in wildlife preservation, aviation security, and defense operations. Reliable identification not only reduces the risk of bird strikes and monitors potential drone threats but also fosters the development of a secure and intelligent ecological environment. However, achieving high precision in recognition under complex and dynamic conditions remains a significant hurdle. To overcome this, we present an advanced detection framework, YOLOv9-AKConv-AFF-GSConv(YOLOv9-AAG), which substantially improves the accuracy of distinguishing birds from drones, even in varied poses, and enhances generalization capabilities in both infrared and visible light scenarios. The framework incorporates several key advancements. First, the GSConv is embedded into the head of the RepNSCPELAN4-AKConv module, enabling efficient extraction of edge and contour information without introducing additional computational cost. Second, the inclusion of the Attentional Feature Fusion (AFF) mechanism within AKConv amplifies the representation of shape features, particularly under challenging environments and multimodal scenarios. Lastly, the optimized AKConv module is effectively integrated with the original RepNSCPELAN4 model, facilitating more accurate and robust feature extraction for small targets, including birds and drones. Experimental evaluations underscore the effectiveness of the proposed framework. On an optical dataset containing multiple bird and drone targets, YOLOv9-AAG achieved recognition accuracy rates of 78.2%, while on an infrared dataset, it reached 86.1%. These results reflect improvements of 7.4% and 6.0%, respectively, over the baseline YOLOv9 model. Additionally, the F1 scores increased by 5.9% and 2.8%. On the publicly available CUB-200-2011 dataset and a combined dataset of visible and infrared images, the model attained mean Average Precision (mAP) scores of 81.9% and 81.6%, respectively, outperforming cutting-edge approaches, including YOLOv9, YOLOv10, and YOLOv11. Furthermore, It was also tested on two videos of drones and birds, the experimental resultsdemonstrated an average detection accuracy of approximately 72% for small bird targets and 80% for drones, highlighting the model’s practical capability to effectively distinguish between these two classes.