LEAF-YOLO: Lightweight Edge-Real-Time Small Object Detection on Aerial Imagery
Advances in Unmanned Aerial Vehicles (UAVs) and deep learning have spotlighted the challenges of detecting small objects in UAV imagery, where limited computational resources complicate deployment on edge devices. While many high-accuracy deep learning solutions have been developed, their large para...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-03-01
|
| Series: | Intelligent Systems with Applications |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2667305325000109 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832540420713742336 |
|---|---|
| author | Van Quang Nghiem Huy Hoang Nguyen Minh Son Hoang |
| author_facet | Van Quang Nghiem Huy Hoang Nguyen Minh Son Hoang |
| author_sort | Van Quang Nghiem |
| collection | DOAJ |
| description | Advances in Unmanned Aerial Vehicles (UAVs) and deep learning have spotlighted the challenges of detecting small objects in UAV imagery, where limited computational resources complicate deployment on edge devices. While many high-accuracy deep learning solutions have been developed, their large parameter sizes hinder deployment on edge devices where low latency and efficient resource use are essential. To address this, we propose LEAF-YOLO, a lightweight and efficient object detection algorithm with two versions: LEAF-YOLO (standard) and LEAF-YOLO-N (nano). Using Lightweight-Efficient Aggregating Fusion along with other blocks and techniques, LEAF-YOLO enhances multiscale feature extraction while reducing complexity, targeting small object detection in dense and varied backgrounds. Experimental results show that both LEAF-YOLO and LEAF-YOLO-N outperform models with fewer than 20 million parameters in accuracy and efficiency on the Visdrone2019-DET-val dataset, running in real-time (>30 FPS) on the Jetson AGX Xavier. LEAF-YOLO-N achieves 21.9% AP.50:.95 and 39.7% AP.50 with only 1.2M parameters. LEAF-YOLO achieves 28.2% AP.50:.95 and 48.3% AP.50 with 4.28M parameters. Furthermore, LEAF-YOLO attains 23% AP.50 on the TinyPerson dataset, outperforming models with ≥ 20 million parameters, making it suitable for UAV-based human detection. |
| format | Article |
| id | doaj-art-c522adcb201944289d07b075013b6d67 |
| institution | Kabale University |
| issn | 2667-3053 |
| language | English |
| publishDate | 2025-03-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Intelligent Systems with Applications |
| spelling | doaj-art-c522adcb201944289d07b075013b6d672025-02-05T04:32:48ZengElsevierIntelligent Systems with Applications2667-30532025-03-0125200484LEAF-YOLO: Lightweight Edge-Real-Time Small Object Detection on Aerial ImageryVan Quang Nghiem0Huy Hoang Nguyen1Minh Son Hoang2Hanoi University of Science and Technology, Hanoi, 10000, Viet NamCorresponding author.; Hanoi University of Science and Technology, Hanoi, 10000, Viet NamHanoi University of Science and Technology, Hanoi, 10000, Viet NamAdvances in Unmanned Aerial Vehicles (UAVs) and deep learning have spotlighted the challenges of detecting small objects in UAV imagery, where limited computational resources complicate deployment on edge devices. While many high-accuracy deep learning solutions have been developed, their large parameter sizes hinder deployment on edge devices where low latency and efficient resource use are essential. To address this, we propose LEAF-YOLO, a lightweight and efficient object detection algorithm with two versions: LEAF-YOLO (standard) and LEAF-YOLO-N (nano). Using Lightweight-Efficient Aggregating Fusion along with other blocks and techniques, LEAF-YOLO enhances multiscale feature extraction while reducing complexity, targeting small object detection in dense and varied backgrounds. Experimental results show that both LEAF-YOLO and LEAF-YOLO-N outperform models with fewer than 20 million parameters in accuracy and efficiency on the Visdrone2019-DET-val dataset, running in real-time (>30 FPS) on the Jetson AGX Xavier. LEAF-YOLO-N achieves 21.9% AP.50:.95 and 39.7% AP.50 with only 1.2M parameters. LEAF-YOLO achieves 28.2% AP.50:.95 and 48.3% AP.50 with 4.28M parameters. Furthermore, LEAF-YOLO attains 23% AP.50 on the TinyPerson dataset, outperforming models with ≥ 20 million parameters, making it suitable for UAV-based human detection.http://www.sciencedirect.com/science/article/pii/S2667305325000109Aerial imageryUAV imagerySmall object detectionEdge-real-time algorithmYou only look once (YOLO) |
| spellingShingle | Van Quang Nghiem Huy Hoang Nguyen Minh Son Hoang LEAF-YOLO: Lightweight Edge-Real-Time Small Object Detection on Aerial Imagery Intelligent Systems with Applications Aerial imagery UAV imagery Small object detection Edge-real-time algorithm You only look once (YOLO) |
| title | LEAF-YOLO: Lightweight Edge-Real-Time Small Object Detection on Aerial Imagery |
| title_full | LEAF-YOLO: Lightweight Edge-Real-Time Small Object Detection on Aerial Imagery |
| title_fullStr | LEAF-YOLO: Lightweight Edge-Real-Time Small Object Detection on Aerial Imagery |
| title_full_unstemmed | LEAF-YOLO: Lightweight Edge-Real-Time Small Object Detection on Aerial Imagery |
| title_short | LEAF-YOLO: Lightweight Edge-Real-Time Small Object Detection on Aerial Imagery |
| title_sort | leaf yolo lightweight edge real time small object detection on aerial imagery |
| topic | Aerial imagery UAV imagery Small object detection Edge-real-time algorithm You only look once (YOLO) |
| url | http://www.sciencedirect.com/science/article/pii/S2667305325000109 |
| work_keys_str_mv | AT vanquangnghiem leafyololightweightedgerealtimesmallobjectdetectiononaerialimagery AT huyhoangnguyen leafyololightweightedgerealtimesmallobjectdetectiononaerialimagery AT minhsonhoang leafyololightweightedgerealtimesmallobjectdetectiononaerialimagery |