YOLO-S: A Lightweight and Accurate YOLO-like Network for Small Target Detection in Aerial Imagery
Small target detection is still a challenging task, especially when looking at fast and accurate solutions for mobile or edge applications. In this work, we present YOLO-S, a simple, fast, and efficient network. It exploits a small feature extractor, as well as skip connection, via both bypass and c...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-02-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/23/4/1865 |
| _version_ | 1797618274521841664 |
|---|---|
| author | Alessandro Betti Mauro Tucci |
| author_facet | Alessandro Betti Mauro Tucci |
| author_sort | Alessandro Betti |
| collection | DOAJ |
| description | Small target detection is still a challenging task, especially when looking at fast and accurate solutions for mobile or edge applications. In this work, we present YOLO-S, a simple, fast, and efficient network. It exploits a small feature extractor, as well as skip connection, via both bypass and concatenation, and a reshape-passthrough layer to promote feature reuse across network and combine low-level positional information with more meaningful high-level information. Performances are evaluated on AIRES, a novel dataset acquired in Europe, and VEDAI, benchmarking the proposed YOLO-S architecture with four baselines. We also demonstrate that a transitional learning task over a combined dataset based on DOTAv2 and VEDAI can enhance the overall accuracy with respect to more general features transferred from COCO data. YOLO-S is from 25% to 50% faster than YOLOv3 and only 15–25% slower than Tiny-YOLOv3, outperforming also YOLOv3 by a 15% in terms of accuracy (mAP) on the VEDAI dataset. Simulations on SARD dataset also prove its suitability for search and rescue operations. In addition, YOLO-S has roughly 90% of Tiny-YOLOv3’s parameters and one half FLOPs of YOLOv3, making possible the deployment for low-power industrial applications. |
| first_indexed | 2024-03-11T08:11:47Z |
| format | Article |
| id | doaj.art-5af83d9276e64579b96e0ae6e7351aac |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-03-11T08:11:47Z |
| publishDate | 2023-02-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-5af83d9276e64579b96e0ae6e7351aac2023-11-16T23:07:13ZengMDPI AGSensors1424-82202023-02-01234186510.3390/s23041865YOLO-S: A Lightweight and Accurate YOLO-like Network for Small Target Detection in Aerial ImageryAlessandro Betti0Mauro Tucci1FlySight srl, via A. Lampredi 45, 57121 Livorno, ItalyDepartment of Energy, Systems, Territory and Construction Engineering, University of Pisa, 56122 Pisa, ItalySmall target detection is still a challenging task, especially when looking at fast and accurate solutions for mobile or edge applications. In this work, we present YOLO-S, a simple, fast, and efficient network. It exploits a small feature extractor, as well as skip connection, via both bypass and concatenation, and a reshape-passthrough layer to promote feature reuse across network and combine low-level positional information with more meaningful high-level information. Performances are evaluated on AIRES, a novel dataset acquired in Europe, and VEDAI, benchmarking the proposed YOLO-S architecture with four baselines. We also demonstrate that a transitional learning task over a combined dataset based on DOTAv2 and VEDAI can enhance the overall accuracy with respect to more general features transferred from COCO data. YOLO-S is from 25% to 50% faster than YOLOv3 and only 15–25% slower than Tiny-YOLOv3, outperforming also YOLOv3 by a 15% in terms of accuracy (mAP) on the VEDAI dataset. Simulations on SARD dataset also prove its suitability for search and rescue operations. In addition, YOLO-S has roughly 90% of Tiny-YOLOv3’s parameters and one half FLOPs of YOLOv3, making possible the deployment for low-power industrial applications.https://www.mdpi.com/1424-8220/23/4/1865aerial imageryconvolutional neural networkvehicle detectionfeature fusionreshape pass-through layercomputer vision |
| spellingShingle | Alessandro Betti Mauro Tucci YOLO-S: A Lightweight and Accurate YOLO-like Network for Small Target Detection in Aerial Imagery Sensors aerial imagery convolutional neural network vehicle detection feature fusion reshape pass-through layer computer vision |
| title | YOLO-S: A Lightweight and Accurate YOLO-like Network for Small Target Detection in Aerial Imagery |
| title_full | YOLO-S: A Lightweight and Accurate YOLO-like Network for Small Target Detection in Aerial Imagery |
| title_fullStr | YOLO-S: A Lightweight and Accurate YOLO-like Network for Small Target Detection in Aerial Imagery |
| title_full_unstemmed | YOLO-S: A Lightweight and Accurate YOLO-like Network for Small Target Detection in Aerial Imagery |
| title_short | YOLO-S: A Lightweight and Accurate YOLO-like Network for Small Target Detection in Aerial Imagery |
| title_sort | yolo s a lightweight and accurate yolo like network for small target detection in aerial imagery |
| topic | aerial imagery convolutional neural network vehicle detection feature fusion reshape pass-through layer computer vision |
| url | https://www.mdpi.com/1424-8220/23/4/1865 |
| work_keys_str_mv | AT alessandrobetti yolosalightweightandaccurateyololikenetworkforsmalltargetdetectioninaerialimagery AT maurotucci yolosalightweightandaccurateyololikenetworkforsmalltargetdetectioninaerialimagery |