GMS-YOLO11n: A Sheep Detection Model for Challenging Fixed-View Farm Conditions Integrating Spatially Gated Structural Enhancement and Multi-Scale Attention
Aug 2026· Animals· Vol 16, pp. 2351· 0 citations· 35 references
Medicine
TL;DR
GMS-YOLO11n, a new computer vision model for automatic sheep detection, improves the original YOLO11n model by strengthening important visual features, such as sheep edges, body contours, and local textures, while combining information from different image scales.
Abstract
Simple Summary Sheep detection is an important part of intelligent farm management because it supports automatic counting, individual tracking, behavior observation, and health monitoring. However, accurately detecting sheep can be difficult when animals gather closely together, partially block one another, appear at different distances from the camera, or are recorded under poor lighting and complex backgrounds. To address these challenges, we developed GMS-YOLO11n, a new computer vision model for automatic sheep detection. Our approach improves the original YOLO11n model by strengthening important visual features, such as sheep edges, body contours, and local textures, while combining information from different image scales. In the current within-farm evaluation, the proposed model reduced missed detections and improved bounding-box localization, particularly in the nighttime low-light and obvious-occlusion subsets. These findings are limited to the fixed-view surveillance conditions represented in the present dataset and should not be interpreted as evidence of general performance across different farms, breeds, camera devices, or viewpoints. External validation is required before the model can be considered suitable for broader farm deployment.
An Adaptive and Scalable YOLO model named AS-YOLOR (Adaptive and Scalable YOLO for Rotated object detection), based on the YOLOv8 baseline is proposed, providing a solution with strong practical potential for achieving efficient and high-precision detection of small, rotated objects.
Jin Huang, Juntao Shen, Min Wang et al.· Applied Sciences· 0 citations
Simple Summary Accurate dairy cow detection is important for automatic barn monitoring, but it remains challenging because cattle often appear at different scales, overlap with each other, and are captured under uneven illumination. The model addresses three practical challenges: dynamic multi-scale feature extraction is used to represent cattle with different body sizes and poses, parameter-free attention is introduced to enhance discriminative features under occlusion and low-contrast conditions, and a lightweight shared detection head is designed to reduce model complexity. Experimental results show that DMFRNet improves localization accuracy while maintaining a compact model size. This study provides an efficient detection approach for dairy cow monitoring in complex farming environments.
Remote sensing images are often featured by small object sizes, varying rotation angles, and background interference, which makes oriented object detection (OOD) from these images challenging. Here, an improved YOLO11s model, termed DAGE-YOLO, is proposed to resolve this challenge. Specifically, parallelized patch-aware attention (PPA) and efficient multi-scale attention (EMA) are incorporated into the backbone of the original YOLO11s to preserve features of small objects and improve cross-scale context modeling. Next, the conventional loss is displaced by the Gaussian Wasserstein Distance (GWD) loss. Moreover, an oriented bounding box (OBB) detection head is introduced to allow the model to achieve precise oriented box localization. Experiments on the DOTAv1.5 and HRSC2016 datasets show that DAGEYOLO achieves an mAP50 8.8% and 7.1% higher than that of the baseline model on the two datasets, respectively, and outperforms several mainstream YOLO-based OOD models. Though this DAGE-YOLO model introduces a moderate increase in the computing overhead, it achieves strong detection accuracy and robustness, offering an effective solution for complex OOD tasks in remote sensing.
Small vehicle detection in aerial images is important for intelligent transportation, low-altitude inspection, urban monitoring, and vision-based sensing systems. Vehicles in aerial images often occupy few pixels and are affected by complex backgrounds, shadows, viewpoint changes, weak texture, and similar class appearances, which can cause missed detections and false alarms. To address these issues, this paper proposes SFC-YOLO, an accuracy-enhanced and parameter-efficient framework based on YOLO11n. A Feature Complementary Block (FCB) is placed at the P5/32 high-level feature stage to enhance local-detail and semantic-context compensation; a Dynamic Feature Alignment Upsampling unit (DFAU) is inserted into the first P5-to-P4 upsampling path to improve content-adaptive feature alignment; and a P5-only Hidden-State Attention (HSA) module is used in the final P5 detection branch to strengthen global semantic interaction. Experiments on the VEDAI eight-class vehicle dataset show that SFC-YOLO reduces the parameter count from 2.584 M to 2.466 M while improving the five-run mean Precision from 0.602 to 0.665, Recall from 0.599 to 0.608, and mAP50 from 0.614 to 0.652. Since the GFLOPs increase from 6.3 to 8.4, SFC-YOLO should be interpreted as a parameter-reduced but not FLOP-reduced framework. The main trade-off is improved detection accuracy with fewer trainable parameters at the cost of higher theoretical computation. Five repeated experiments yield an average mAP50 of 0.6523 ± 0.0117, quantifying the run-to-run variation under the current training protocol.
Jinwei Zuo, Qiyi He, Fulin Liu et al.· Italian National Conference...· 0 citations
In rural and forest-bordering areas, human-wildlife conflict has escalated, resulting in substantial crop damage, property loss, and human mortality. Conventional mitigating solutions such as fencing, manual surveillance, and local alarms, have limited coverage, time taking and high operating costs. Automated real-time animal detection systems have become a practical for early warning and conflict avoidance due to recent upgrades in edge computing and deep learning particularly in distant locations without dependable internet access. In this study, embedded hardware and deep learning-based object detection are used to create a real-time wild animal detection and warning system. Roboflow was used to develop and enhance a dataset of 1,554 real-world photographs from seven classes (tiger, bear, elephant, lion, cheetah, leopard, and human) to 5,917 images. Using GPU acceleration, three object detection models YOLOv8(n)(n), SSD, and RetinaNet were trained on Google Colab and assessed using precision, recall, F1-score, and mean Average Precision (mAP). A GSM 900 A module is used to send SMS notification with no requirement of internet connectivity, and the optimized model was installed on a Raspberry Pi 5 with a Pi Camera for real-time inference. According to experimental data, YOLOv8(n) performs better than SSD and RetinaNet, with an mAP@50 84.4% and recall of 81% as opposed to SSD’s 62.3% precision and 58% recall and RetinaNet’s around 77% precision and 80% recall. Additionally, YOLOv8(n) showed improved robustness and quicker inference time across a range of background and lighting circumstances. While SSD and RetinaNet demonstrated weaker or scattered attention in complex situations, explainable AI analysis utilizing Grad-CAM and LIME verified greater and more concentrated activation on pertinent animal traits for YOLOv8(n). For real-time wildlife monitoring and early warning, the suggested system offers a precise, affordable, and scalable option. Through the integration of YOLOv8(n), Raspberry Pi 5 edge computing, and GSM-based connectivity, the system efficiently functions in distant settings and dramatically shortens reaction times during wildlife infiltration incidents. The findings demonstrate that YOLOv8(n) is the best model for integrated animal detection, enhancing wildlife conservation, lowering agricultural loss, and improving human safety.
M. Patil, A. Madankar, Roshan Umate· Discover Computing· 0 citations
Simple Summary The reliable identification of individual cattle is important for precision livestock farming because it supports the long-term monitoring of animal behavior, health, and welfare. However, automatically identifying Hanwoo cattle is challenging because different individuals often have very similar body appearance, and their visual features can change greatly with pose and camera viewpoint. To address this problem, this study proposes a visual re-identification framework that combines pose-guided image generation with viewpoint-aware feature refinement. Specifically, the method learns to generate pose-diverse, identity-consistent images and, during inference, aggregates their features with the original image features to obtain a more compact identity representation, which is further refined using viewpoint information. Experimental results show that the method improves cattle re-identification performance across different datasets and evaluation settings. This work provides a useful technical reference for developing non-invasive, camera-based cattle identification systems in real farm environments.
Jiaqi Liu, A. Fuentes, Shujie Han et al.· Animals· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
