An Improved YOLOv5 Method for Small Object Detection in UAV Capture Scenes

Liu, Zhen; Gao, Xuehui; Wan, Yu; Wang, Jianhao; Lyu, Hao

doi:10.1109/access.2023.3241005

Cited by 55 publications

(26 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Following the path of specialists in machine learning, the study by Lu et al (2023) introduced Yolov8's cuttingedge technology to the study of plants. There were also a few simple yet effective adjustments made.…”

Section: Related Workmentioning

confidence: 99%

Smart Farming with Sooty Tern Optimization based LS-HGNet Classification Model

Krishnan,

Vikranth,

Sumithra

et al. 2024

IJERR

View full text Add to dashboard Cite

Smart farming technologies enable farmers to use resources like water, fertilizer and pesticides as efficiently as possible. This paper discusses how Unmanned Aerial Vehicle (UAV) pictures can be used to automatically detect and count tassels, thereby advancing the advancement of strategic maize planting. The real state of affairs in cornfields is complicated, though, and the current algorithms struggle to provide the speed and accuracy required for real-time detection. This research employed a sizable, excellent dataset of maize tassels to solve this problem. This paper suggests using the bottom-hat-top-hat preprocessing technique to address the lighting irregularities and noise in maize photos taken by drones. The Lightweight weight-stacked hourglass Network (LS-HGNet) model is suggested for classification. The hourglass network structure of LS-HGNet, which is mostly utilised as a backbone network, has allowed significant advancements in the discovery of maize tassels. In light of this, the current work suggests a lighter variant of the hourglass network that also enhances the accuracy of tassel detection in maize plants. The additional skip connections used in the new hourglass network architecture allow minimal changes to the number of network parameters while improving performance. Consequently, the suggested LS-HGNet classifier lowers the computational burden and increases the convolutional receptive field. The hyperparameter tuning process is then carried out using the Sooty Tern Optimisation Algorithm (STOA), which helps increase tassel detection accuracy. Numerous tests were conducted to verify that the suggested approach is more accurate at 98.7% and more efficient than the most advanced techniques currently in use.

show abstract

Section: Related Workmentioning

confidence: 99%

Smart Farming with Sooty Tern Optimization based LS-HGNet Classification Model

Krishnan,

Vikranth,

Sumithra

et al. 2024

IJERR

View full text Add to dashboard Cite

show abstract

“…In the absence of radial distortion, the coordinates (X c , Y c ) in the image coordinates are given by [26][27][28][29][30][31][32]  …”

Section: Ugv Pose Estimationmentioning

confidence: 99%

Vision-Guided Tracking and Emergency Landing for UAVs on Moving Targets

Debele,

Shi,

Wondosen

et al. 2024

Drones

View full text Add to dashboard Cite

This paper presents a vision-based adaptive tracking and landing method for multirotor Unmanned Aerial Vehicles (UAVs), designed for safe recovery amid propulsion system failures that reduce maneuverability and responsiveness. The method addresses challenges posed by external disturbances such as wind and agile target movements, specifically, by considering maneuverability and control limitations caused by propulsion system failures. Building on our previous research in actuator fault detection and tolerance, our approach employs a modified adaptive pure pursuit guidance technique with an extra adaptation parameter to account for reduced maneuverability, thus ensuring safe tracking of moving objects. Additionally, we present an adaptive landing strategy that adapts to tracking deviations and minimizes off-target landings caused by lateral tracking errors and delayed responses, using a lateral offset-dependent vertical velocity control. Our system employs vision-based tag detection to ascertain the position of the Unmanned Ground Vehicle (UGV) in relation to the UAV. We implemented this system in a mid-mission emergency landing scenario, which includes actuator health monitoring of emergency landings. Extensive testing and simulations demonstrate the effectiveness of our approach, significantly advancing the development of safe tracking and emergency landing methods for UAVs with compromised control authority due to actuator failures.

show abstract

“…The YOLO family has many models, but they perform differently on different datasets. YOLOv5 is easy to deploy and train, has good reliability and stability [27]. At the same time, Web of Science shows that in the past year, YOLOv5-based publications have an absolute advantage and are widely used.…”

Section: The Algorithm Principle Of Yolov5mentioning

confidence: 99%

An Improved YOLOv5 Method to Detect Tailings Ponds from High-Resolution Remote Sensing Images

Sun

Meng

et al. 2023

Remote Sensing

View full text Add to dashboard Cite

Tailings ponds’ failure and environmental pollution make tailings monitoring very important. Remote sensing technology can quickly and widely obtain ground information and has become one of the important means of tailings monitoring. However, the efficiency and accuracy of traditional remote sensing monitoring technology have difficulty meeting the management needs. At the same time, affected by factors such as the geographical environment and imaging conditions, tailings have various manifestations in remote sensing images, which all bring challenges to the accurate acquisition of tailings information in large areas. By improving You Only Look Once (YOLO) v5s, this study designs a deep learning-based framework for the large-scale extraction of tailings ponds information from the entire high-resolution remote sensing images. For the improved YOLOv5s, the Swin Transformer is integrated to build the Swin-T backbone, the Fusion Block of efficient Reparameterized Generalized Feature Pyramid Network (RepGFPN) in DAMO-YOLO is introduced to form the RepGFPN Neck, and the head is replaced with Decoupled Head. In addition, sample boosting strategy (SBS) and global non-maximum suppression (GNMS) are designed to improve the sample quality and suppress repeated detection frames in the entire image, respectively. The model test results based on entire Gaofen-6 (GF-6) high-resolution remote sensing images show that the F1 score of tailings ponds is significantly improved by 12.22% compared with YOLOv5, reaching 81.90%. On the basis of both employing SBS, the improved YOLOv5s boots the mAP@0.5 of YOLOv5s by 5.95%, reaching 92.15%. This study provides a solution for tailings ponds’ monitoring and ecological environment management.

show abstract

An Improved YOLOv5 Method for Small Object Detection in UAV Capture Scenes

Cited by 55 publications

References 34 publications

Smart Farming with Sooty Tern Optimization based LS-HGNet Classification Model

Smart Farming with Sooty Tern Optimization based LS-HGNet Classification Model

Vision-Guided Tracking and Emergency Landing for UAVs on Moving Targets

An Improved YOLOv5 Method to Detect Tailings Ponds from High-Resolution Remote Sensing Images

Contact Info

Product

Resources

About