Individual Animal and Herd Identification Using Custom YOLO v3 and v4 with Images Taken from a UAV Camera at Different Altitudes

Petso, Tinao; Jamisola, Rodrigo S.; Mpoeleng, Dimane; Mmereki, Wazha

doi:10.1109/icsip52628.2021.9688827

Cited by 7 publications

(2 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Gupta et al [28] used web mining techniques to develop a custom dataset to identify individual cattle breeds. Petso et al [29] improved the YOLO v3 and v4 models for individual and group detection by capturing wildlife images from drones at different altitudes. Furthermore, Rosli et al [30] proposed YOLOv4 based on a deep learning algorithm for efficient and accurate underwater detection to overcome the aquatic environment's turbidity, dynamic background, and low visibility and improve the underwater vision system.…”

Section: B Yolo Identification Methodsmentioning

confidence: 99%

Open Pose Mask R-CNN Network for Individual Cattle Recognition

Wang,

Zhang,

Gao

et al. 2023

IEEE Access

View full text Add to dashboard Cite

Cattle's individual identification plays a crucial role in effectively managing large farms. To enhance agricultural efficiency, promote the digital transformation of animal husbandry, and improve animal welfare, it is essential to employ advanced identification technologies capable of real-time monitoring of cattle individual. This paper introduces a novel network called Open Pose Mask R-CNN (OP-Mask R-CNN) for individual cattle identification, which combines Open Pose with the Mask R-CNN network. Three key strategies are presented to improve the identification of individual cattle. First, optimize the number of convolutional layers in the Mask R-CNN backbone network, i.e., ResNet101. Second, introduce an Open Pose-based bovine skeleton feature extraction method. Finally, construct a fusion mechanism that combines the attention module, the convolutional block attention module (CBAM), the open pose module, and the ResNet101. Experimental results demonstrate that our proposed method achieves a 5.6% increase in recognition accuracy and improves recognition speed compared to the original Mask R-CNN model. This work strikes a balance between accuracy and complexity, facilitating the development of a lightweight bovine individual recognition technique.

show abstract

Section: B Yolo Identification Methodsmentioning

confidence: 99%

Open Pose Mask R-CNN Network for Individual Cattle Recognition

Wang,

Zhang,

Gao

et al. 2023

IEEE Access

View full text Add to dashboard Cite

show abstract

“…For the identification and detection of herds of white rhinos, giraffes, wildebeests, and zebras, the work of Petso, T., Jamisola, R. S., Mpoeleng, D. and Mmereki, W. (2021) 32 used YOLO-based detectors for detection through images captured by drones. The challenge is to detect animals with aerial images, as there may be animals camouflaged with the environment.…”

Section: Related Workmentioning

confidence: 99%

Evaluating YOLO-based Object Detectors for Detecting Road-Killed Endangered Brazilian Animals

Ferrante

Nakamura

Sampaio

et al. 2023

Preprint

View full text Add to dashboard Cite

Wildlife roadkill is a recurring, dangerous problem that affects both humans and animals and has received increasing attention from environmentalists worldwide. Addressing this problem is difficult due to the high investments required in road infrastructure to effectively reduce wildlife vehicle collisions. Despite recent applications of machine learning techniques in low-cost and economically viable detection systems, e.g., for alerting drivers about the presence of animals and collecting statistics on endangered animal species, the success and wide adoption of these systems depend heavily on the availability of data for system training. The lack of training data negatively impacts the feature extraction of machine learning models, which is crucial for successful animal detection and classification. In this paper, we evaluate the performance of several state-of-the-art object detection models on limited data for model training. The selected models are based on the YOLO architecture, which is well-suited for and commonly used in real-time object detection. These include the YoloV4, Scaled-YoloV4, YoloV5, YoloR, YoloX, and YoloV7 models. We focus on Brazilian endangered animal species and use the BRA-Dataset for model training. We also assess the effectiveness of data augmentation and transfer learning techniques in our evaluation. The models are compared using summary metrics such as precision, recall, mAP, and FPS and are qualitatively analyzed considering classic computer vision problems. The results show that the architecture with the best results against false negatives is Scaled-YoloV4, while the best FPS detection score is the nano version of YoloV5.

show abstract

A Review on Deep Learning on UAV Monitoring Systems for Agricultural Applications

Petso

Jamisola

2023

Studies in Computational Intelligence

View full text Add to dashboard Cite

Individual Animal and Herd Identification Using Custom YOLO v3 and v4 with Images Taken from a UAV Camera at Different Altitudes

Cited by 7 publications

References 22 publications

Open Pose Mask R-CNN Network for Individual Cattle Recognition

Open Pose Mask R-CNN Network for Individual Cattle Recognition

Evaluating YOLO-based Object Detectors for Detecting Road-Killed Endangered Brazilian Animals

A Review on Deep Learning on UAV Monitoring Systems for Agricultural Applications

Contact Info

Product

Resources

About