AI and Related Technologies in the Fields of Smart Agriculture: A Review

Information, Journal Year: 2025, Volume and Issue: 16(2), P. 100 - 100

Published: Feb. 2, 2025

The integration of cutting-edge technologies—such as the Internet Things (IoT), artificial intelligence (AI), machine learning (ML), and various emerging technologies—is revolutionizing agricultural practices, enhancing productivity, sustainability, efficiency. objective this study is to review literature regarding development evolution AI well other technologies in fields Agriculture they are developed transformed by integrating above technologies. areas examined open field smart farming, vertical indoor zero waste agriculture, precision livestock greenhouses, regenerative agriculture. This paper links current research, technological innovations, case studies present a comprehensive these being context for benefit farmers consumers general. By exploring practical applications future perspectives, work aims provide valuable insights address global food security challenges, minimize environmental impacts, support sustainable goals through application new

Language: Английский

---

Benchmarking wild bird detection in complex forest scenes

Ecological Informatics, Journal Year: 2024, Volume and Issue: 80, P. 102466 - 102466

Published: Jan. 9, 2024

Camera traps are widely used for wildlife monitoring and making informed conservation land-management decisions, but the resulting 'big data' laborious to process. Deep learning-based methods have been adopted detection in camera traps. However, these detect large mammals uncomplicated scenes, where powerful deep-learning models work effectively. Few studies conducted develop artificial intelligence recognizing wild birds that live complicated field scenes with protective colors small sizes. Here we a dataset of 9717 images from 15 bird species based on test 8 object algorithms (Faster RCNN, Cascade RetinaNet, FCOS, RepPoints, ATSS, Deformable-DETR, Sparse RCNN) assess their performance. We also explored effect different backbones model accuracy. Among them, RCNN performs best, mAP 0.693 capabilities. Models perform differently certain species, significantly affect accuracy model. utilizing Swin-T backbone is best-performing combination, 0.704. This study could help researchers identify efficiently inspires research recognition complex ecological settings.

Language: Английский

---

Research trends in livestock facial identification – A review

Journal of Animal Science and Technology, Journal Year: 2025, Volume and Issue: 67(1), P. 43 - 55

Published: Jan. 1, 2025

This review examines the application of video processing and convolutional neural network (CNN)-based deep learning for animal face recognition, identification, re-identification. These technologies are essential precision livestock farming, addressing challenges in production efficiency, welfare, environmental impact. With advancements computer technology, monitoring systems have evolved into sensor-based contact methods video-based non-contact methods. Recent developments enable continuous analysis accumulated data, automating conditions. By integrating with CNN-based learning, it is possible to estimate growth, identify individuals, monitor behavior more effectively. enhance management systems, leading improved outcomes, sustainability farming practices.

Language: Английский

---

Improved Wildlife Recognition through Fusing Camera Trap Images and Temporal Metadata

Diversity, Journal Year: 2024, Volume and Issue: 16(3), P. 139 - 139

Published: Feb. 23, 2024

Camera traps play an important role in biodiversity monitoring. An increasing number of studies have been conducted to automatically recognize wildlife camera trap images through deep learning. However, recognition by alone is often limited the size and quality dataset. To address above issues, we propose Temporal-SE-ResNet50 network, which aims improve accuracy exploiting temporal information attached images. First, constructed SE-ResNet50 network extract image features. Second, obtained metadata from images, after cyclical encoding, used a residual multilayer perceptron (MLP) obtain Finally, features were fused identification dynamic MLP module. The experimental results on Camdeboo dataset show that fusing about 93.10%, improvement 0.53%, 0.94%, 1.35%, 2.93%, 5.98%, respectively, compared with ResNet50, VGG19, ShuffleNetV2-2.0x, MobileNetV3-L, ConvNeXt-B models. Furthermore, demonstrate effectiveness proposed method different national park datasets. Our provides new idea for animal domain knowledge further recognition, can better serve conservation ecological research.

Language: Английский

---

Development of Apple Detection System and Reinforcement Learning for Apple Manipulator

Electronics, Journal Year: 2023, Volume and Issue: 12(3), P. 727 - 727

Published: Feb. 1, 2023

Modern deep learning systems make it possible to develop increasingly intelligent solutions in various fields of science and technology. The electronics single board computers facilitate the control robotic solutions. At same time, implementation such tasks does not require a large amount resources. However, models still high level computing power. Thus, effective an robot manipulator is when computationally complex model on GPU graphics devices mechanics unit single-board computer work together. In this regard, study devoted development vision for estimation coordinates objects interest, as well subsequent recalculation relative form action. addition, simulation environment, reinforcement was developed determine optimal path picking apples from 2D images. detection efficiency test images 92%, laboratory achieve 100% apples. algorithm has been trained that provides adequate guidance located at distance 1 m along Z axis. original neural network used recognize using big image dataset, algorithms estimating were investigated, use suggested optimize policy.

Language: Английский

---

How to achieve accurate wildlife detection by using vehicle‐mounted mobile monitoring images and deep learning?

Remote Sensing in Ecology and Conservation, Journal Year: 2025, Volume and Issue: unknown

Published: March 14, 2025

Abstract With the advancement of artificial intelligence (AI) technologies, vehicle‐mounted mobile monitoring systems have become increasingly integrated into wildlife practices. However, images captured through these often present challenges such as low resolution, small target sizes, and partial occlusions. Consequently, detecting animal targets using conventional deep‐learning networks is challenging. To address challenges, this paper presents an enhanced YOLOv7 model, referred to YOLOv7(sr‐sm), which incorporates a super‐resolution (SR) reconstruction module object optimization module. The YOLOv7(sr‐sm) model introduces that leverages generative adversarial (GANs) reconstruct high‐resolution details from blurry images. Additionally, attention mechanism Neck Head form module, enhances model's ability detect locate densely packed targets. Using system, four taxa—sheep, birds, deer, antelope —were on Tibetan Plateau. These were combined with publicly available photographs create test dataset. Experiments conducted dataset, comparing eight popular detection models. results demonstrate significant improvements in precision, recall, mean Average Precision (mAP), achieving 93.9%, 92.1%, 92.3%, respectively. Furthermore, compared newly released YOLOv8l outperforms it by 9.3%, 2.1%, 4.5% three metrics while also exhibiting superior parameter efficiency higher inference speeds. architecture can accurately identify images, serving reliable tool for identification counting systems. findings provide technological support application intelligent techniques biodiversity conservation efforts.

Language: Английский

---

Deep learning for Amur tiger re-identification in camera traps: A tool assisting population monitoring and spatio-temporal analysis

Ecological Indicators, Journal Year: 2025, Volume and Issue: 171, P. 113227 - 113227

Published: Feb. 1, 2025

Language: Английский

---

Automated Animal Intrusion Detection: A Deep Learning Approach

Lecture notes in electrical engineering, Journal Year: 2025, Volume and Issue: unknown, P. 195 - 208

Published: Jan. 1, 2025

Language: Английский

---

YOLO-Animal: An efficient wildlife detection network based on improved YOLOv5

2022 International Conference on Image Processing, Computer Vision and Machine Learning (ICICML), Journal Year: 2022, Volume and Issue: unknown, P. 464 - 468

Published: Oct. 28, 2022

With the continuous development of modern society, rapid expansion human civilization squeezes living space other organisms, and extinction more biological species has sounded alarm for us. Therefore, in order to timely understand changes wild animals resources a specific area, facilitate relevant personnel formulate effective restoration protection measures, this paper proposes Animal recognition network based on deep learning improved YOLOv5: YOLO-Animal. The application artificial intelligence computer vision covered all aspects human's daily life work. benchmark YOLOv5 algorithm can quickly accurately deal with problems associated images. Through fusion weighted Bidirectional Feature Pyramid Network (BiFPN) Effective Channel Attention (ECA) module, original YOLOv5s structure is enhanced, detection accuracy small targets occluded fuzzy effectively improved. YOLO-Animal model outperforms by 3.2% mAP achieves 95.5% test set.

Language: Английский

---

A systematic study on transfer learning: Automatically identifying empty camera trap images using deep convolutional neural networks

Ecological Informatics, Journal Year: 2024, Volume and Issue: 80, P. 102527 - 102527

Published: Feb. 17, 2024

Transfer learning is extensively utilized for automatically recognizing and filtering out empty camera trap images that lack animal presence. Current research uses transfer identifying typically solely updates the fully connected layer of models, they usually select a pre-trained source model only based on its relevance to target task. However, do not consider optimization update selection, nor investigate effect sample size class number domain data set used construct performance model. Both these are issues worth exploring. We answered two using three different datasets ResNext-101 Our experimental results showed when 20,000 training samples from ImageNet dataset Snapshot Serengeti dataset, our proposed optimal layers improved accuracy 92.9% 95.5% (z = −7.087, p < 0.001, N 8118) compared existing method updating layer. A similar improvement was observed transferring Lasha Mountain dataset. Additionally, indicated increasing binary-class build 100,000 1 million, 90.4% 93.5% −3.869, 8948). Similar were obtained constructing ten classifications. Based results, we drew following conclusions: (1) instead commonly can significantly improve model's performance. (2) The varied transferred same (3) classes in did impact positively correlated with performance, there might be threshold effect.

Language: Английский

---