Computer Vision Models Research Articles

Real-time digital dermatitis detection in dairy cows on Android and iOS apps using computer vision techniques

Abstract The aim of the study was to deploy computer vision (CV) models for real-time detection of digital dermatitis (DD) lesions in cows using Android or iOS mobile applications. Early detection of DD lesions in dairy cows is crucial for prompt treatment and animal welfare. Android and iOS apps could facilitate routine and early DD detection in cows’ feet on dairy and beef farms. Upon detecting signs of DD, dairy farmers could implement preventive and treatment methods, including foot baths, topical treatment, hoof trimming, or quarantining cows affected by DD to prevent its spread. We applied transfer-learning to DD image data for 5 lesion classes, M0, M4H, M2, M2P, and M4P, on pretrained YOLOv5 model architecture using COCO-128 pretrained weights. The combination of localization loss, classification loss, and objectness loss was used for the optimization of prediction performance. The custom DD detection model was trained on 363 images of size 416 x 416 pixels and tested on 46 images. During model training, data were augmented to increase model robustness in different environments. The model was converted into TFLite format for Android devices and CoreML format for iOS devices. Techniques such as quantization were implemented to improve inference speed in real-world settings. The DD models achieved an average mAP (mean Average Precision) of 0.95 on the test dataset. When tested in real-time, iOS devices resulted in Cohen’s kappa value of 0.57 (95% CI: 0.49-0.65) averaged across the 5 lesion classes denoting the moderate agreement of the model detection with human investigators. The Android device resulted in a Cohen’s kappa value of 0.38 (95% CI: 0.29-0.47) denoting fair agreement between model and investigator. Combining M2 and M2P classes and M4H and M4P classes resulted in a Cohen’s kappa value of 0.65 (95% CI: 0.54-0.76) and 0.46 (95% CI: 0.35-0.57), for Android and iOS devices, respectively. For the two-class model (lesion vs. non-lesion), a Cohen’s kappa value of 0.74 (95% CI: 0.63-0.85) and 0.65 (95% CI: 0.52-0.78) was achieved for iOS and Android devices, respectively. iOS achieved a good inference time of 20 ms, compared to 57 ms on Android. Additionally, we deployed models on Ultralytics iOS and Android apps giving kappa scores of 0.56 (95% CI: 0.48-0.64) and 0.46 (95% CI: 0.37-0.55) respectively. Our custom iOS app surpassed the Ultralytics apps in terms of kappa score and the confidence score.

Integration of Deep Learning with Fox Optimization Algorithm for Early Detection and Classification of Tomato Leaf and Fruit Diseases

Tomato is a common vegetable crop extensively cultivated in the farming lands in India. The hot climate of India is perfect for its development, but particular weather conditions along with many other aspects affect the growing of tomato plants. Apart from these natural disasters and weather conditions, plant diseases consist a major issue in crop production. Precisely classifying leaf and fruit diseases in tomato plants is a vital step toward computerizing processes. Traditional disease detection models for tomato crops often fall short in predictability. To address this, Machine Learning (ML) and Deep Learning (DL) models have been developed, presenting advanced classification capabilities and the ability to manage the vast variability in agricultural data that conventional computer vision models struggle with. This work presents an Integration of DL with Fox Optimization Algorithm (FOA) for the Recognition and Classification of Tomato Leaf and Fruit Diseases (IDLFOA-DCTLFD). The major objective of the proposed IDLFOA-DCTLFD model is to enhance the detection and classification outcomes of tomato leaf and fruit diseases. At the initial stage, the Median Filter (MF) model is used for pre-processing and the Efficient Channel Attention-SqueezeNet (ECA-SqueezeNet) model is employed for feature extraction. For the hyperparameter tuning process, the proposed IDLFOA-DCTLFD technique implements the FOA. Finally, a Wasserstein Generative Adversarial Network (WGAN) is utilized for the detection of tomato leaf and fruit diseases. The IDLFOA-DCTLFD method is experimentally examined in a tomato leaf and fruit dataset. The experimental validation of the IDLFOA-DCTLFD methodology portrayed a superior accuracy value of 98.02%, surpassing the existing techniques.

A Comprehensive Review on Biomedical Image Classification using Deep Learning Models

Medical imaging is one of the most efficient tools for visualizing the interior organs of the body and its associated diseases. Medical imaging is used to diagnose diseases and offer treatment. Since the manual examination of a massive number of Medical Images (MI) is a laborious and erroneous task, automated MI analysis approaches have been developed for computer-aided diagnostic solutions to reduce time and enhance diagnostic quality. Deep Learning (DL) models have exhibited excellent performance in the MI segmentation, classification, and detection process. This article presents a comprehensive review of the recently developed DL-based MIK classification models for various diseases. The current review aims to assist researchers and physicians of biomedical imaging in understanding the basic concepts and recent DL models. It explores recent MI classification techniques developed for various diseases. A thorough discussion on Computer Vision (CV) and DL models is also carried out.

Real-Time Driver Drowsiness Detection Using Facial Analysis and Machine Learning Techniques

Drowsy driving poses a significant challenge to road safety worldwide, contributing to thousands of accidents and fatalities annually. Despite advancements in driver drowsiness detection (DDD) systems, many existing methods face limitations such as intrusiveness and delayed reaction times. This research addresses these gaps by leveraging facial analysis and state-of-the-art machine learning techniques to develop a real-time, non-intrusive DDD system. A distinctive aspect of this research is its systematic assessment of various machine and deep learning algorithms across three pivotal public datasets, the NTHUDDD, YawDD, and UTA-RLDD, known for their widespread use in drowsiness detection studies. Our evaluation covered techniques including the K-Nearest Neighbors (KNNs), support vector machines (SVMs), convolutional neural networks (CNNs), and advanced computer vision (CV) models such as YOLOv5, YOLOv8, and Faster R-CNN. Notably, the KNNs classifier reported the highest accuracy of 98.89%, a precision of 99.27%, and an F1 score of 98.86% on the UTA-RLDD. Among the CV methods, YOLOv5 and YOLOv8 demonstrated exceptional performance, achieving 100% precision and recall with mAP@0.5 values of 99.5% on the UTA-RLDD. In contrast, Faster R-CNN showed an accuracy of 81.0% and a precision of 63.4% on the same dataset. These results demonstrate the potential of our system to significantly enhance road safety by providing proactive alerts in real time.

A Novel Approach to Road Safety: Detecting Illegal Overtaking Using Smartphone Cameras and Deep Learning for Vehicle Auditing

Overtaking relies heavily on the driver’s attention and cognitive state, and illegal overtaking can lead to accidents, severe injuries, or fatalities. To enhance highway safety, we propose a method for accurately detecting illegal overtaking on continuous road lanes. We used dashboard-mounted smartphone cameras and geolocation data to filter the analysis areas. We used the state-of-the-art deep learning model You Only Look Once version 8 (YOLOv8) to detect yellow road lanes. When these lanes suggest potential illegal overtaking, we apply the YOLO for Panoptic driving Perception version 2 (YOLOPv2) model, followed by post-processing. We confirm overtaking events by checking for overlaps between detections from both models. We store confirmed instances and evaluate the information temporally rather than just from individual frames. We then analyze the entire video to identify violations and extract the moments of occurrence. We tested the algorithm on real-world traffic data under various weather and lighting conditions. Our method demonstrates reliability and consistency in identifying illegal overtaking. We achieved 16 TP and only 1 FP over 56 videos totaling 41 h, 9 min, and 24 s, with precision, recall, and F1-score values of 1.000, 0.941, and 0.970, respectively. Consequently, our innovative and practical solution, utilizing simple cameras and advanced computer vision models, can significantly enhance highway safety and support vehicle auditing systems.

Deep learning and computer vision in plant disease detection: a comprehensive review of techniques, models, and trends in precision agriculture

Plant diseases cause significant damage to agriculture, leading to substantial yield losses and posing a major threat to food security. Detection, identification, quantification, and diagnosis of plant diseases are crucial parts of precision agriculture and crop protection. Modernizing agriculture and improving production efficiency are significantly affected by using computer vision technology for crop disease diagnosis. This technology is notable for its non-destructive nature, speed, real-time responsiveness, and precision. Deep learning (DL), a recent breakthrough in computer vision, has become a focal point in agricultural plant protection that can minimize the biases of manually selecting disease spot features. This study reviews the techniques and tools used for automatic disease identification, state-of-the-art DL models, and recent trends in DL-based image analysis. The techniques, performance, benefits, drawbacks, underlying frameworks, and reference datasets of more than 278 research articles were analyzed and subsequently highlighted in accordance with the architecture of computer vision and deep learning models. Key findings include the effectiveness of imaging techniques and sensors like RGB, multispectral, and hyperspectral cameras for early disease detection. Researchers also evaluated various DL architectures, such as convolutional neural networks, vision transformers, generative adversarial networks, vision language models, and foundation models. Moreover, the study connects academic research with practical agricultural applications, providing guidance on the suitability of these models for production environments. This comprehensive review offers valuable insights into the current state and future directions of deep learning in plant disease detection, making it a significant resource for researchers, academicians, and practitioners in precision agriculture.

PHARAOH: A collaborative crowdsourcing platform for phenotyping and regional analysis of histology

Deep learning has proven capable of automating key aspects of histopathologic analysis. However, its context-specific nature and continued reliance on large expert-annotated training datasets hinders the development of a critical mass of applications to garner widespread adoption in clinical/research workflows. Here, we present an online collaborative platform that streamlines tissue image annotation to promote the development and sharing of custom computer vision models for PHenotyping And Regional Analysis Of Histology (PHARAOH; https://www.pathologyreports.ai/). Specifically, PHARAOH uses a weakly supervised, human-in-the-loop learning framework whereby patch-level image features are leveraged to organize large swaths of tissue into morphologically-uniform clusters for batched annotation by human experts. By providing cluster-level labels on only a handful of cases, we show how custom PHARAOH models can be developed efficiently and used to guide the quantification of cellular features that correlate with molecular, pathologic and patient outcome data. Moreover, by using our PHARAOH pipeline, we showcase how correlation of cohort-level cytoarchitectural features with accompanying biological and outcome data can help systematically devise interpretable morphometric models of disease. Both the custom model design and feature extraction pipelines are amenable to crowdsourcing, positioning PHARAOH to become a fully scalable, systems-level solution for the expansion, generalization and cataloging of computational pathology applications.

A human single-neuron dataset for object recognition

Object recognition is fundamental to how we interact with and interpret the world around us. The human amygdala and hippocampus play a key role in object recognition, contributing to both the encoding and retrieval of visual information. Here, we recorded single-neuron activity from the human amygdala and hippocampus when neurosurgical epilepsy patients performed a one-back task using naturalistic object stimuli. We employed two sets of naturalistic object images from leading datasets extensively used in primate neural recordings and computer vision models: we recorded 1204 neurons using the ImageNet stimuli, which included broader object categories (10 different images per category for 50 categories), and we recorded 512 neurons using the Microsoft COCO stimuli, which featured a higher number of images per category (50 different images per category for 10 categories). Together, our extensive dataset, offering the highest spatial and temporal resolution currently available in humans, will not only facilitate a comprehensive analysis of the neural correlates of object recognition but also provide valuable opportunities for training and validating computational models.

Deep generative modeling of annotated bacterial biofilm images

Biofilms are critical for understanding environmental processes, developing biotechnology applications, and progressing in medical treatments of various infections. Nowadays, a key limiting factor for biofilm analysis is the difficulty in obtaining large datasets with fully annotated images. This study introduces a versatile approach for creating synthetic datasets of annotated biofilm images with employing deep generative modeling techniques, including VAEs, GANs, diffusion models, and CycleGAN. Synthetic datasets can significantly improve the training of computer vision models for automated biofilm analysis, as demonstrated with the application of Mask R-CNN detection model. The approach represents a key advance in the field of biofilm research, offering a scalable solution for generating high-quality training data and working with different strains of microorganisms at different stages of formation. Terabyte-scale datasets can be easily generated on personal computers. A web application is provided for the on-demand generation of biofilm images.

The assessment of bactericidal efficacy of seasidewater samples and automated detection of sulfate-reducing bacteria using computer vision models

The assessment of bactericidal efficacy of seasidewater samples and automated detection of sulfate-reducing bacteria using computer vision models

Flower Longevity Quantification in Greenhouses Using Deep Learning Models for Computer Vision

Flower Longevity Quantification in Greenhouses Using Deep Learning Models for Computer Vision

Graphical User Interface for Handwritten Mathematical Expression Employing RNN-based Encoder-decoder Model

Aim: Scientific, technical, and educational research domains all heavily rely on handwritten mathematical expressions. The extensive use of online handwritten mathematical expression recognition is a consequence of the availability of strong computational touchscreen appliances, such as the recent development of deep neural networks as superior sequence recognition models. Background: Further investigation and enhancement of these technologies are vital to tackle the contemporary obstacles presented by the widespread adoption of remote learning and work arrangements as a result of the global health crisis. Objective: Handwritten document processing has gained more attention in the last ten years due to notable developments in deep neural network-based computer vision models and sequence recognition, as well as the widespread proliferation of touch and pen-enabled smartphones and tablets. It comes naturally to people to write by hand in daily interactions. Method: In this patent article, authors implemented Hand written expressions using RNNbased encoder for the CROHME dataset. Later, the proposed model was validated using CNNbased encoder and End-to-end encoder decoder techniques. The proposed model is also validated on other datasets. Results: The RNN-based encoder model yields 82.78%, while the CNN-based encoder model and end-to-end encoder-decoder technique yield 81.38% and 80.73%, respectively. Conclusion: 1.6% accuracy improvement was attained over CNN-based encoder while 2.4% accuracy improvement over end-to-end encoder-decoder. CROHME dataset 2019 version results in better accuracy than other datasets.

An image and video dataset of nesting green sea turtles with annotated data

Photo- and video-based reidentification of green sea turtles using their natural markers is far less invasive than artificial tagging. An RGB camera mounted on a man-portable rig, was used to collect video data on Greater Talang Island (1 °54’45″N 109 °46’33″E) from September to October 2022, and September 2023. This islet is located 30 minutes offshore from the Sematan district in Southwest Sarawak, Malaysia. In total, 42 videos of 42 unique individuals were captured, from which 14,471 video frames were extracted and annotated with a bounding box each. Through manual inspection, the annotation data were diligently checked and edited if necessary. Additionally, 130 selected frames were further annotated each with a region-of-interest mask containing only the carapace. These data have the potential to aid researchers in training computer vision models for various tasks such as counting, segmentation and individual reidentification.

Leveraging ordinal depth for accurate transmission estimation in single image dehazing

ABSTRACT Real-time images with haze do not meet the input quality standards for computer vision models. Existing dehazing algorithms have shown effective results at the nearby pixels. However, they fail to dehaze the pixels as the distance regions. Also, some of the models fail when the group of pixels contains white patches such as clouds. This paper proposed an approach that not only clears haze in nearby regions but also distant sky regions. The proposed model estimates image transmission based on the scene's depth information. The ordinal depth of the scene is first estimated and categorized as foreground, background, and sky. Using the transmission values of the foreground pixels, the transmission values for the other pixels are recalculated and applied to the Atmospheric Scattering model to produce the dehazed image. Extensive testing on datasets has shown that the proposed model delivers superior results compared to existing state-of-the-art image dehazing networks.

UAV (Unmanned Aerial Vehicle): Diverse Applications of UAV Datasets in Segmentation, Classification, Detection, and Tracking

Unmanned Aerial Vehicles (UAVs) have transformed the process of data collection and analysis in a variety of research disciplines, delivering unparalleled adaptability and efficacy. This paper presents a thorough examination of UAV datasets, emphasizing their wide range of applications and progress. UAV datasets consist of various types of data, such as satellite imagery, images captured by drones, and videos. These datasets can be categorized as either unimodal or multimodal, offering a wide range of detailed and comprehensive information. These datasets play a crucial role in disaster damage assessment, aerial surveillance, object recognition, and tracking. They facilitate the development of sophisticated models for tasks like semantic segmentation, pose estimation, vehicle re-identification, and gesture recognition. By leveraging UAV datasets, researchers can significantly enhance the capabilities of computer vision models, thereby advancing technology and improving our understanding of complex, dynamic environments from an aerial perspective. This review aims to encapsulate the multifaceted utility of UAV datasets, emphasizing their pivotal role in driving innovation and practical applications in multiple domains.

AI Techniques for Plant Disease Detection

Plant diseases affect agricultural production, food security, and economic stability, making them a major concern for global agriculture. To reduce losses and guarantee sustainable farming methods, these diseases must be identified early and managed effectively. Manual inspections, which are labour-intensive, unreliable, and unscalable for large-scale agricultural applications, are frequently the basis of traditional disease monitoring techniques. Innovative approaches to plant disease tracking have been made possible by the quick development of artificial intelligence (AI), which offers improved scalability, accuracy, and efficiency. This study provides a comprehensive assessment of AI-based plant disease tracking systems, concentrating on techniques that integrate machine learning, deep learning, computer vision, and data-driven models. Key uses include integrating satellite imagery and IoT-enabled devices for real-time monitoring, predicting disease outbreaks using environmental data, and detecting diseases using picture processing. Additionally examined is the function of mobile applications in providing farmers with easily accessible diagnostic tools. The study also discusses important issues like model generalisation, data scarcity, computational constraints, and socioeconomic obstacles to AI adoption in agriculture. This review highlights the revolutionary potential of AI in building resilient agricultural systems, ultimately promoting global food security and sustainable development, by combining recent developments and pointing out research needs

Real -Time Network Packet Classification Exploiting Computer Vision Architectures

The upcoming 6G and NextG networks underscore the necessity of sophisticated security methods based on Artificial Intelligence (AI) in order to detect malicious activity and adjust to new threats. Because computer vision techniques may be used to recognize complex patterns, their incorporation into the cybersecurity industry is a promising development. In this work, we present a computationally effective categorization technique that enforces the real-time conversion of packets into pictures by directly acting upon the raw packets gathered at base stations. The suggested solution's novel features include its lightweight implementation, which well satisfies the requirements of upcoming 6G networks, and its network edge operation, which permits early threat detection as near to the packet origin as feasible. We examine the efficacy of this methodology in terms of F1-score and prediction time by employing cutting-edge computer vision architectures and a customized Convolutional Neural Network (CNN) to tackle an intrusion detection task utilizing a substantial 5G dataset. The CNN design is superior than complicated models, as demonstrated by the results of experiments. The CNN consistently beats the other cutting-edge computer vision models over several packet window sizes N (i.e., 10, 50, and 100 packets), reaching very high F1-scores (0.99593, 0.99860, and 0.99895). A scalability investigation reveals a trade- off between the performance and scalability of CNN, with higher N values resulting in longer prediction times. However, the scalability of the other computer vision models is superior, allowing for an ideal model selection free of compromises

Digitizing the trails; the design of a low-cost wearable camera rig and development of a computer vision tree species detection model

Abstract. This paper presents a novel solution to a reality capture and 3D contextual mapping problem. A wearable camera rig was designed to capture photographic information of existing trails within North Vancouver, British Columbia, Canada. High-resolution video and 360° images were captured for 3D reconstruction models and the creation of an online virtual tour environment. Pathway width of the existing trails was extracted using a linear feature extraction tool to create detailed base map drawings for the client. A computer vision model was trained to detect tree species from images of tree bark. A summary of the stages of the project, lessons learnt from the use of photogrammetric (Structure from Motion) methods, suggested improvements to the computer vision model, and thoughts on the direction of reality capture methods within the infrastructure sector are given.

Developing A Fast Computer Vision Model for Diagnosing and Classifying Hip Fractures

Developing A Fast Computer Vision Model for Diagnosing and Classifying Hip Fractures

Automated Assessment of Pelvic Longitudinal Rotation Using Computer Vision in Canine Hip Dysplasia Screening.

Canine hip dysplasia (CHD) screening relies on accurate positioning in the ventrodorsal hip extended (VDHE) view, as even mild pelvic rotation can affect CHD scoring and impact breeding decisions. This study aimed to assess the association between pelvic rotation and asymmetry in obturator foramina areas (AOFAs) and to develop a computer vision model for automated AOFA measurement. In the first part, 203 radiographs were analyzed to examine the relationship between pelvic rotation, assessed through asymmetry in iliac wing and obturator foramina widths (AOFWs), and AOFAs. A significant association was found between pelvic rotation and AOFA, with AOFW showing a stronger correlation (R2 = 0.92, p < 0.01). AOFW rotation values were categorized into minimal (n = 71), moderate (n = 41), marked (n = 37), and extreme (n = 54) groups, corresponding to mean AOFA ± standard deviation values of 33.28 ± 27.25, 54.73 ± 27.98, 85.85 ± 41.31, and 160.68 ± 64.20 mm2, respectively. ANOVA and post hoc testing confirmed significant differences in AOFA across these groups (p < 0.01). In part two, the dataset was expanded to 312 images to develop the automated AOFA model, with 80% allocated for training, 10% for validation, and 10% for testing. On the 32 test images, the model achieved high segmentation accuracy (Dice score = 0.96; Intersection over Union = 0.93), closely aligning with examiner measurements. Paired t-tests indicated no significant differences between the examiner and model's outputs (p > 0.05), though the Bland-Altman analysis identified occasional discrepancies. The model demonstrated excellent reliability (ICC = 0.99) with a standard error of 17.18 mm2. A threshold of 50.46 mm2 enabled effective differentiation between acceptable and excessive pelvic rotation. With additional training data, further improvements in precision are expected, enhancing the model's clinical utility.