Junhao Dai

Mr. Junhao Dai


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
17 August 2024 in World Electric Vehicle Journal
Verified
Journal Article
Open access
PDF file
KZKai Zhu
Kai Zhu
JDJunhao Dai
Junhao Dai
ZGZhenchao Gu
Zhenchao Gu
The development of electric vehicles has facilitated intelligent transportation, which requires the swift and effective detection and tracking of moving vehicles. To satisfy this demand, this paper presents an enhanced DeepSORT algorithm. By selecting YOLO-SSFS as the front-end detector and incorporating a lightweight and high-precision feature training network called FasterNet, the proposed method effectively extracts vehicle appearance attributes. Besides this, the noise scale adaptive Kalman filter is implemented and the conventional cascade matching process is substituted with global join matching, thereby enhancing overall performance and tracking accuracy. Validation conducted on the VisDrone dataset demonstrates the superiority of this method compared to the original DeepSORT algorithm, exhibiting a 4.76% increase in tracking accuracy and a 3.10% improvement in tracking precision. The findings reveal the advantages of the algorithms in the domain of vehicle detection and tracking, allowing significant technological advancements in intelligent transportation systems.

7 January 2026 in Preprints.org
Verified
Preprint
Open access
PDF file
JDJunhao Dai
Junhao Dai
KZKai Zhu
Kai Zhu
Infrared traffic object detection faces challenges such as low resolution, weak thermal 2 contrast, and inefficiency in detecting small objects. To address these issues, this paper 3 proposes RES-YOLO, an enhanced YOLOv8n-based architecture. It incorporates Receptive 4 Field Adaptive Convolution for improved multi-scale perception, Efficient Multi-scale 5 Attention for better feature representation, and the Scylla-IoU loss for more accurate 6 and faster bounding box regression. Additionally, a pseudo-color infrared dataset is 7 constructed to enrich texture and contrast information beyond conventional white-hot 8 images. Experiments on both the FLIR public dataset and a self-built dataset show RES- 9 YOLO improves accuracy by 4.9% and 5.5% over the baseline while maintaining real-time 10 performance. These results highlight the method’s effectiveness in integrating lightweight 11 deep learning and dataset enhancement for robust perception in intelligent vehicle systems, 12 supporting AI-driven autonomous driving and driver assistance applications.