TPH-YOLOv5: Improved YOLOv5 Based on Transformer Prediction Head for Object Detection on Drone-captured Scenarios
TPH-YOLOv5: Улучшенный YOLOv5 с трансформерной головой предсказания для обнаружения объектов на снимках с дронов
2021-10-01
SCID: 54.1/eqx7xsjk
Discuss with AI
CBAM (Convolutional Block Attention Module)TPH-YOLOv5Transformer Prediction HeadVisDrone2021 datasetYOLOv5
Figures from the paper
Abstract (AI)
Object detection on drone-captured scenarios is a recent popular task. As drones always navigate in different altitudes, the object scale varies violently, which burdens the optimization of networks. Moreover, high-speed and low-altitude flight bring in the motion blur on the densely packed objects, which leads to great challenge of object distinction. To solve the two issues mentioned above, we propose TPH-YOLOv5. Based on YOLOv5, we add one more prediction head to detect different-scale objects. Then we replace the original prediction heads with Transformer Prediction Heads (TPH) to explore the prediction potential with self-attention mechanism. We also integrate convolutional block attention model (CBAM) to find attention region on scenarios with dense objects. To achieve more improvement of our proposed TPH-YOLOv5, we provide bags of useful strategies such as data augmentation, multi-scale testing, multi-model integration and utilizing extra classifier. Extensive experiments on dataset VisDrone2021 show that TPH-YOLOv5 have good performance with impressive interpretability on drone-captured scenarios. On DET-test-challenge dataset, the AP result of TPH-YOLOv5 are 39.18%, which is better than previous SOTA method (DPNetV3) by 1.81%. On VisDrone Challenge 2021, TPH-YOLOv5 wins 5thplace and achieves well-matched results with 1stplace model (AP 39.43%). Compared to baseline model (YOLOv5), TPH-YOLOv5 improves about 7%, which is encouraging and competitive.
Key Findings
1
Added an extra prediction head to YOLOv5 to better detect objects across violently varying scales in drone imagery.
2
Applied training and inference strategies (data augmentation, multi-scale testing, multi-model integration, extra classifier) to further boost performance.
3
Integrated CBAM (convolutional block attention module) to better locate attention regions in densely packed, motion-blurred drone scenes.
4
On VisDrone2021 DET-test-challenge, TPH-YOLOv5 achieved AP 39.18%, outperforming prior SOTA DPNetV3 by 1.81% and improving ≈7% over baseline YOLOv5; placed 5th in VisDrone Challenge 2021 and matched the 1st-place model (AP 39.43%).
5
Replaced original prediction heads with Transformer Prediction Heads (TPH) using self-attention to improve prediction capability.
Research Object
YOLOv5-based object detection model (TPH-YOLOv5) applied to drone-captured imagery
Research Subject
Improving detection of multi-scale, densely packed and motion-blurred objects in drone-captured scenarios via an added prediction head, Transformer Prediction Heads (self-attention), CBAM attention, and related training/testing strategies to increase AP
Publication Details
Publication Date
2021-10-01
Journal
Publisher
ISSN
Cited by
2463
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest
References available in scid.ai7
Deep Residual Learning for Image Recognition2016
Very Deep Convolutional Networks for Large-Scale Image Recognition2014
Exploiting Generative AI to Scale up Intelligent Tutoring Systems2023
Swin Transformer: Hierarchical Vision Transformer using Shifted Windows2021
An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale2020
mixup: Beyond Empirical Risk Minimization2017
A Multi-Modal Distributed Real-Time IoT System for Urban Traffic Control (Invited Paper)2024