We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Indoor Location Mapping of Lameness Chickens with Multi Cameras and Perspective Transform Using Convolutional Neural Networks.
- Authors
Wiwit Agus Triyanto; Adi, Kusworo; Suseno, Jatmiko Endro
- Abstract
Lameness is one of the most serious diseases affecting chickens, which can also increase the risk of premature culling of chickens and cause huge economic losses. So far, the process of detecting chicken lameness and finding out its location is still carried out traditionally by farmers checking directly in the cage, but this can actually result in increased stress levels in the chickens. Computer vision-based approaches with deep learning have been widely used to help farm automation, but there are several things that need to be considered and are problems; these include light variables, occlusion. In this study, Faster Regions with Convolutional Neural Network (Faster R-CNN), Single Shot MultiBox Detector (SSD) and You Only Look Once (YOLO), which is a Convolutional Neural Network (CNN) network model was chosen to perform the detection, tracking, and mapping of chicken locations. YOLOv8 was combined Adam Optimizer to improve training performance. Based on the results, customized YOLOv8 has the best mAP, support, precision and F1-Score values compared to the others, with 0.922, 0.987, 0.990 and 0.988. The matrix of transformation and coordinate-to-meter conversion produces chicken locations that match real conditions, not just the position of pixel (x, y) coordinates. From the detection and tracking, the location of 1 sick (lameness) chicken and 7 healthy chickens were obtained. The results of this research can properly display the movement and position of chickens in the cage, so they can be used to monitor chicken welfare.
- Subjects
CONVOLUTIONAL neural networks; CHICKENS; DEEP learning; CHICKEN diseases; CAMERAS
- Publication
Mathematical Modelling of Engineering Problems, 2024, Vol 11, Issue 2, p543
- ISSN
2369-0739
- Publication type
Article
- DOI
10.18280/mmep.110227