Faster R-CNN Research Articles

Evaluating Machine Learning-Based Elephant Recognition in Complex African Landscapes Using Drone Imagery

Abstract This paper evaluates a machine learning-based approach for identifying and analyzing African bush elephants within complex terrains using high-resolution drone imagery. With human-wildlife conflict posing a significant threat to elephants worldwide, accurate and efficient monitoring techniques are crucial, yet challenging in diverse landscapes. Our study utilizes approximately 3,180 drone-captured images from Kasungu National Park in Malawi, encompassing various terrains including dense forests and open bushlands. These images were systematically preprocessed and analyzed using three distinct ML algorithms: Faster R-CNN, RetinaNet, and Mask R-CNN, each fine-tuned for identification of elephants across different age groups. Comparative performance metrics revealed nuanced strengths and limitations: Faster R-CNN showed notable proficiency in detecting adult elephants, particularly in dense foliage. Mask R-CNN, while less precise overall, demonstrated increased effectiveness in identifying juveniles and infants. RetinaNet, optimized for larger images, showed particular adeptness with adult elephants but less so with younger ones. Despite these promising results, overall recognition rates were lower than ideal, highlighting the complexities of wildlife identification in natural settings. This study not only facilitates the identification and counting of individual elephants but also provides insights into the challenges of applying ML in complex ecological contexts. The derived insights can assist conservationists and park officials in making informed decisions related to wildlife protection and habitat preservation. Furthermore, the study offers a valuable blueprint for integrating AI and machine learning technology into wildlife conservation strategies, presenting a scalable model with potential applications for different species and geographic regions, while acknowledging the need for further refinement to enhance accuracy and reliability in diverse ecological settings.

Study on intelligent recognition of urban road subgrade defect based on deep learning.

China's operational highway subgrades exhibit a trend of diversifying types and an increasing number of defects, leading to more frequent urban road safety incidents. This paper starts from the non-destructive testing of urban road subgrade defects using geological radar, aiming to achieve intelligent identification of subgrade pathologies with geological radar. The GprMax forward simulation software is used to establish multi-layer composite structural models of the subgrade, studying the characteristics of geological radar images for different types of subgrade diseases. Based on the forward simulation images of geological radar for subgrade defects and field measurement data, a geological radar subgrade defect image database is established. The Faster R-CNN deep learning algorithm is applied to achieve target detection, recognition, and classification of subgrade defect images. By comparing the loss value, total number of identified regions, and recognition accuracy as metrics, the study compares four improved versions of the Faster R-CNN algorithm. The results indicate that the faster_rcnn_inception_v2 version is more suitable for the intelligent identification of non-destructive testing of urban road subgrade defects.

A two-stage deep-learning model for determination of the contact of mandibular third molars with the mandibular canal on panoramic radiographs.

This study aimed to assess the accuracy of a two-stage deep learning (DL) model for (1)detecting mandibular third molars (MTMs) and the mandibular canal (MC), and (2)classifying the anatomical relationship between these structures (contact/no contact) on panoramic radiographs. MTMs and MCs were labeled on panoramic radiographs by a calibrated examiner using bounding boxes. Each bounding box contained MTM and MC on one side. The relationship of MTMs with the MC was assessed on CBCT scans by two independent examiners without the knowledge of the condition of MTM and MC on the corresponding panoramic image, and dichotomized as contact/no contact. Data were split into training, validation, and testing sets with a ratio of 80:10:10. Faster R-CNN was used for detecting MTMs and MCs and ResNeXt for classifying their relationship. AP50 and AP75 were used as outcomes for detecting MTMs and MCs, and accuracy, precision, recall, F1-score, and the area-under-the-receiver-operating-characteristics curve (AUROC) were used to assess classification performance. The training and validation of the models were conducted using the Python programming language with the PyTorch framework. Three hundred eighty-seven panoramic radiographs were evaluated. MTMs were present bilaterally on 232 and unilaterally on 155 radiographs. In total, 619 images were collected which included MTMs and MCs. AP50 and AP75 indicating accuracy for detecting MTMs and MCs were 0.99 and 0.90 respectively. Classification accuracy, recall, specificity, F1-score, precision, and AUROC values were 0.85, 0.85, 0.93, 0.84, 0.86, and 0.91, respectively. DL can detect MTMs and MCs and accurately assess their anatomical relationship on panoramic radiographs.

Cucumber Leaf Segmentation Based on Bilayer Convolutional Network

When monitoring crop growth using top-down images of the plant canopies, leaves in agricultural fields appear very dense and significantly overlap each other. Moreover, the image can be affected by external conditions such as background environment and light intensity, impacting the effectiveness of image segmentation. To address the challenge of segmenting dense and overlapping plant leaves under natural lighting conditions, this study employed a Bilayer Convolutional Network (BCNet) method for accurate leaf segmentation across various lighting environments. The major contributions of this study are as follows: (1) Utilized Fully Convolutional Object Detection (FCOS) for plant leaf detection, incorporating ResNet-50 with the Convolutional Block Attention Module (CBAM) and Feature Pyramid Network (FPN) to enhance Region of Interest (RoI) feature extraction from canopy top-view images. (2) Extracted the sub-region of the RoI based on the position of the detection box, using this region as input for the BCNet, ensuring precise segmentation. (3) Utilized instance segmentation of canopy top-view images using BCNet, improving segmentation accuracy. (4) Applied the Varifocal Loss Function to improve the classification loss function in FCOS, leading to better performance metrics. The experimental results on cucumber canopy top-view images captured in glass greenhouse and plastic greenhouse environments show that our method is highly effective. For cucumber leaves at different growth stages and under various lighting conditions, the Precision, Recall and Average Precision (AP) metrics for object recognition are 97%, 94% and 96.57%, respectively. For instance segmentation, the Precision, Recall and Average Precision (AP) metrics are 87%, 83% and 84.71%, respectively. Our algorithm outperforms commonly used deep learning algorithms such as Faster R-CNN, Mask R-CNN, YOLOv4 and PANet, showcasing its superior capability in complex agricultural settings. The results of this study demonstrate the potential of our method for accurate recognition and segmentation of highly overlapping leaves in diverse agricultural environments, significantly contributing to the application of deep learning algorithms in smart agriculture.

A Study to Investigate the Role and Challenges Associated with the Use of Deep Learning in Autonomous Vehicles

The application of deep learning in autonomous vehicles has surged over the years with advancements in technology. This research explores the integration of deep learning algorithms into autonomous vehicles (AVs), focusing on their role in perception, decision-making, localization, mapping, and navigation. It shows how deep learning, as a part of machine learning, mimics the human brain’s neural networks, enabling advancements in perception, decision-making, localization, mapping, and overall navigation. Techniques like convolutional neural networks are used for image detection and steering control, while deep learning is crucial for path planning, automated parking, and traffic maneuvering. Localization and mapping are essential for AVs’ navigation, with deep learning-based object detection mechanisms like Faster R-CNN and YOLO proving effective in real-time obstacle detection. Apart from the roles, this study also revealed that the integration of deep learning in AVs faces challenges such as dataset uncertainty, sensor challenges, and model training intricacies. However, these issues can be addressed through the increased standardization of sensors and real-life testing for model training, and advancements in model compression technologies can optimize the performance of deep learning in AVs. This study concludes that deep learning plays a crucial role in enhancing the safety and reliability of AV navigation. This study contributes to the ongoing discourse on the optimal integration of deep learning in AVs, aiming to foster their safety, reliability, and societal acceptance.

Comparison of CNN-Based Architectures for Detection of Different Object Classes

(1) Background: Detecting people and technical objects in various situations, such as natural disasters and warfare, is critical to search and rescue operations and the safety of civilians. A fast and accurate detection of people and equipment can significantly increase the effectiveness of search and rescue missions and provide timely assistance to people. Computer vision and deep learning technologies play a key role in detecting the required objects due to their ability to analyze big volumes of visual data in real-time. (2) Methods: The performance of the neural networks such as You Only Look Once (YOLO) v4-v8, Faster R-CNN, Single Shot MultiBox Detector (SSD), and EfficientDet has been analyzed using COCO2017, SARD, SeaDronesSee, and VisDrone2019 datasets. The main metrics for comparison were mAP, Precision, Recall, F1-Score, and the ability of the neural network to work in real-time. (3) Results: The most important metrics for evaluating the efficiency and performance of models for a given task are accuracy (mAP), F1-Score, and processing speed (FPS). These metrics allow us to evaluate both the accuracy of object recognition and the ability to use the models in real-world environments where high processing speed is important. (4) Conclusion: Although different neural networks perform better on certain types of metrics, YOLO outperforms them on all metrics, showing the best results of mAP-0.88, F1-0.88, and FPS-48, so the focus was on these models.

Multi-class Object Detection in Urban Scenes Based on Deep Learning

Abstract. The rapid development of urbanization presents challenges and requirements for multi-class object detection in urban scenes. Accurately identifying buildings, vehicles, and trees in urban scenes can optimize urban planning, traffic management, monitoring environmental conditions, and promote the development of smart cities. Traditional target detection methods perform poorly in complex urban environments, while deep learning technology achieves accurate target recognition and positioning by automatically extracting high-level semantic features. In this study, we chose to use the YOLOv5s algorithm for multi-class target detection in urban scenes. YOLOv5s is a lightweight deep learning model with small storage space and efficient detection speed. In this paper, the Potsdam area data published by ISPRS is used to make the label data of buildings, vehicles and trees. The YOLOv5s algorithm is used to iteratively train the model. The results show that the mAP value detected by the YOLOv5s model can reach 82.83%. The experimental results show that the algorithm shows higher accuracy than SSD and Faster R-CNN in tree detection. Although it has a slight decline in building and vehicle detection, considering the factors such as detection accuracy, speed, and model size, the YOLOv5s algorithm has a better recognition and detection effect for the detection of multi-class targets in urban scenes.

Comparative performance of YOLOv8, YOLOv9, YOLOv10, YOLOv11 and Faster R-CNN models for detection of multiple weed species

Weeds pose a serious production challenge in various agronomic crops by reducing their grain yields. Increasing cases of herbicide-resistant (HR) weed populations further exacerbate the problem. Future weed control tactics require the integration of non-chemical and reduced chemical-based strategies that can target site- and specie-specific weed management (SSSWM). Advanced machine learning technology has the potential to localize and detect weed seedlings to implement SSSWM. However, due to large biological variability among various weed species and environmental conditions where they grow, accurate and precise weed detection remains challenging. The main objectives of this research were to (1) develop an annotated image database of cocklebur (Xanthium strumarium L.), dandelion (Taraxacum officinale), common waterhemp (Amaranthus tuberculatus), Palmer amaranth (Amaranthus palmeri) and common lambsquarters (Chenopodium album L.), and (2) investigate the comparative performance (speed and accuracy) of YOLOv8, YOLOv9, YOLOv10, YOLOv11 and Faster R-CNN algorithms in detecting those weed species. A weed dataset with bounding box annotations for each weed species was created, consisting of images collected under variable field conditions which were preprocessed and augmented to create a dataset of 2348 color images. The YOLOv8, YOLOv9, YOLOv10, YOLOv11 and Faster R-CNN were trained using the annotated weed image database to detect each weed species. Results indicated that YOLOv11 was the fastest model with inference time of 13.5 milliseconds (ms) followed by YOLOv8 and YOLOv10 with inference time 23 and 19.3 milliseconds (ms), respectively. The YOLOv9 had the highest accuracy in detecting different weed species with an overall mean average precision (mAP@0.5) of 0.935. In contrast, the Detectron2 with Fast R-CNN configuration provided mAP@0.5 of 0.821 with an inference time of 63.8 ms. These results suggest that the YOLO series algorithms have the potential for real-time deployment for weed species detection more accurately and faster than Faster R-CNN in agricultural fields.

Enhancing physician support in pancreatic cancer diagnosis: New M-F-RCNN artificial intelligence model using endoscopic ultrasound.

Background and study aims Endoscopic ultrasound (EUS) is vital for early pancreatic cancer diagnosis. Advances in artificial intelligence (AI), especially deep learning, have improved medical image analysis. We developed and validated the Modified Faster R-CNN (M-F-RCNN), an AI algorithm using EUS images to assist in diagnosing pancreatic cancer. Methods We collected EUS images from 155 patients across three endoscopy centers from July 2022 to July 2023. M-F-RCNN development involved enhancing feature information through data preprocessing and utilizing an improved Faster R-CNN model to identify cancerous regions. Its diagnostic capabilities were validated against an external set of 1,000 EUS images. In addition, five EUS doctors participated in a study comparing the M-F-RCNN model's performance with that of human experts, assessing diagnostic skill improvements with AI assistance. Results Internally, the M-F-RCNN model surpassed traditional algorithms with an average precision of 97.35%, accuracy of 96.49%, and recall rate of 5.44%. In external validation, its sensitivity, specificity, and accuracy were 91.7%, 91.5%, and 91.6%, respectively, outperforming non-expert physicians. The model also significantly enhanced the diagnostic skills of doctors. Conclusions: The M-F-RCNN model shows exceptional performance in diagnosing pancreatic cancer via EUS images, greatly improving diagnostic accuracy and efficiency, thus enhancing physician proficiency and reducing diagnostic errors.

The Detection and Classification of Scaphoid Fractures in Radiograph by Using a Convolutional Neural Network.

Scaphoid fractures, particularly occult and non-displaced fractures, are difficult to detect using traditional X-ray methods because of their subtle appearance and variability in bone density. This study proposes a two-stage CNN approach to detect and classify scaphoid fractures using anterior-posterior (AP) and lateral (LA) X-ray views for more accurate diagnosis. This study emphasizes the use of multi-view X-ray images (AP and LA views) to improve fracture detection and classification. The multi-view fusion module helps integrate information from both views to enhance detection accuracy, particularly for occult fractures that may not be visible in a single view. The proposed method includes two stages, which are stage 1: detect the scaphoid bone using Faster RCNN and a Feature Pyramid Network (FPN) for region proposal and small object detection. The detection accuracy for scaphoid localization is 100%, with Intersection over Union (IoU) scores of 0.8662 for AP views and 0.8478 for LA views. And stage 2: perform fracture classification using a ResNet backbone and FPN combined with a multi-view fusion module to combine features from both AP and LA views. This stage achieves a classification accuracy of 89.94%, recall of 87.33%, and precision of 90.36%. The proposed model performs well in both scaphoid bone detection and fracture classification. The multi-view fusion approach significantly improves recall and accuracy in detecting fractures compared to single-view approaches. In scaphoid detection, both AP and LA views achieved 100% detection accuracy. In fracture detection, using multi-view fusion, the accuracy for AP views reached 87.16%, and for LA views, it reached 83.83%. The multi-view fusion model effectively improves the detection of scaphoid fractures, particularly in cases of occult and non-displaced fractures. The model provides a reliable, automated approach to assist clinicians in detecting and diagnosing scaphoid fractures more efficiently.

MS-YOLO: A Lightweight and High-Precision YOLO Model for Drowning Detection.

A novel detection model, MS-YOLO, is developed in this paper to improve the efficiency of drowning rescue operations. The model is lightweight, high in precision, and applicable for intelligent hardware platforms. Firstly, the MD-C2F structure is built to capture the subtle movements and posture changes in various aquatic environments, with a light weight achieved by introducing dynamic convolution (DcConv). To make the model perform better in small object detection, the EMA mechanism is incorporated into the MD-C2F. Secondly, the MSI-SPPF module is constructed to improve the performance in identifying the features of different scales and the understanding of complex backgrounds. Finally, the ConCat single-channel fusion is replaced by BiFPN weighted channel fusion to retain more feature information and remove the irrelevant information in drowning features. Relative to the Faster R-CNN, SSD, YOLOv6, YOLOv9, and YOLOv10, the MS-YOLO achieves an average accuracy of 86.4% in detection on a self-built dataset at an ultra-low computational cost of 7.3 GFLOPs.

TableExtractNet: A Model of Automatic Detection and Recognition of Table Structures from Unstructured Documents

This paper presents TableExtractNet, a model that automatically finds and understands tables from scanned documents, tasks that are essential for quick use of information in many fields. This is driven by the growing need for efficient and accurate table interpretation in business documents where tables enhance data communication and aid decision-making. The model uses a mix of two advanced techniques, CornerNet and Faster R-CNN, to accurately locate tables and understand their layout. Tests on standard datasets, IIIT-AR-13K, STDW, SciTSR, and PubTabNet, show that this model performs better than previous ones, making it very good at dealing with tables that have complicated designs or are in documents with a lot of detail. The success of this model marks a step forward in making document analysis more automated. It makes it easier to turn complex scanned documents containing tables into data that are more manipulable by computers.

Deep learning networks-based tomato disease and pest detection: a first review of research studies using real field datasets.

Recent advances in deep neural networks in terms of convolutional neural networks (CNNs) have enabled researchers to significantly improve the accuracy and speed of object recognition systems and their application to plant disease and pest detection and diagnosis. This paper presents the first comprehensive review and analysis of deep learning approaches for disease and pest detection in tomato plants, using self-collected field-based and benchmarking datasets extracted from real agricultural scenarios. The review shows that only a few studies available in the literature used data from real agricultural fields such as the PlantDoc dataset. The paper also reveals overoptimistic results of the huge number of studies in the literature that used the PlantVillage dataset collected under (controlled) laboratory conditions. This finding is consistent with the characteristics of the dataset, which consists of leaf images with a uniform background. The uniformity of the background images facilitates object detection and classification, resulting in higher performance-metric values for the models. However, such models are not very useful in agricultural practice, and it remains desirable to establish large datasets of plant diseases under real conditions. With some of the self-generated datasets from real agricultural fields reviewed in this paper, high performance values above 90% can be achieved by applying different (improved) CNN architectures such as Faster R-CNN and YOLO.

Detection of carotid plaques on panoramic radiographs using deep learning

ObjectivesPanoramic radiographs (PRs) can reveal an incidental finding of atherosclerosis, or carotid artery calcification (CAC), in 3-15% of examined patients. However, limited training in identification of such calcifications among dental professionals results in missed diagnoses. This study aimed to detect CAC on PRs using an artificial intelligence (AI) model based on a vision transformer. Methods6,404 PRs were obtained from one hospital and screened for the presence of CAC based on electronic medical records. CAC was manually annotated with bounding boxes by an oral radiologist and reviewed and revised by three experienced clinicians to achieve consensus. An AI approach based on Faster R-CNN and Swin Transformer was trained and evaluated based on 185 PRs with CAC and 185 PRs without CAC. Reported and replicated diagnostic performances of published AI approaches based on convolutional neural networks (CNNs) were used for comparison. Quantitative evaluation of the performance of the models included precision, F1-score, recall, area-under-the-curve (AUC), and average precision (AP). ResultsThe proposed method based on Faster R-CNN and Swin Transformer achieved a precision of 0.895, recall of 0.881, F1-score of 0.888, AUC of 0.950, and AP of 0.942, surpassing models based on a CNN. ConclusionsThe detection performance of this newly developed and validated model was improved compared to previously reported models. Clinical significanceIntegrating AI models into dental imaging to assist dental professionals in the detection of CAC on PRs has the potential to significantly enhance the early detection of carotid artery atherosclerosis and its clinical management.

Survivor detection approach for post earthquake search and rescue missions based on deep learning inspired algorithms

Rapid and reliable detection of human survivors trapped under debris is crucial for effective post-earthquake search and rescue (SAR) operations. This paper presents a novel approach to survivor detection using a snake robot equipped with deep learning (DL) based object identification algorithms. We evaluated the performance of three main algorithms: Faster R-CNN, Single Shot MultiBox Detector (SSD), and You Only Look Once (YOLO). While these algorithms are initially trained on the PASCAL VOC 2012 dataset for human identification, we address the lack of a dedicated dataset for trapped individuals by compiling a new dataset of 200 images that specifically depicts this scenario, featuring cluttered environment and occlusion. Our evaluation takes into account detection accuracy, confidence interval, and running time. The results demonstrate that the YOLOv10 algorithm achieves the 98.4 mAP@0.5, accuracy of 98.5% for inference time of 15 ms. We validate the performance of these algorithms using images of human survivors trapped under debris and subjected to various occlusions.

Hybrid Deep Learning Based Model for Removing Grid-Line Artifacts from Radiographical Images

The digital imaging technique known as Computed Radiography (CR) has transformed the medical imaging industry by providing a number of advantages. It eliminates the need for traditional film-based methods, making it more efficient and convenient. A common issue faced with CR images is the presence of grid artifacts and other pattern artifacts, which can have a significant impact on the quality of the images when viewed on a computer screen, especially if a clinic-grade display is not accessible. This paper presents a novel framework for removing grid line artifacts from X-ray images, which is a critical challenge in medical imaging. The framework proposes a hybrid Deep Grid model that combines a Gaussian band-stop filter with ADAM optimization to produce high-quality, grid-line free X-ray images that are suitable for further analysis and diagnosis. Deep learning (DL) models for instance the Convolutional Neural Network (CNN), DenseNet, VGG-Net, and Fast R-CNN were utilized to classify images, and the grid-by-grid removal of grid lines in the image was performed. The proposed framework achieved a high accuracy rate of 98% in eliminating grid line artifacts from X-ray images, demonstrating its possibility for a big improvement the accuracy and reliability of diagnostics for medical based on X-ray images

Investigation of Unsafe Construction Site Conditions Using Deep Learning Algorithms Using Unmanned Aerial Vehicles.

The rapid adoption of Unmanned Aerial Vehicles (UAVs) in the construction industry has revolutionized safety, surveying, quality monitoring, and maintenance assessment. UAVs are increasingly used to prevent accidents caused by falls from heights or being struck by falling objects by ensuring workers comply with safety protocols. This study focuses on leveraging UAV technology to enhance labor safety by monitoring the use of personal protective equipment, particularly helmets, among construction workers. The developed UAV system utilizes the tensorflow technique and an alert system to detect and identify workers not wearing helmets. Employing the high-precision, high-speed, and widely applicable Faster R-CNN method, the UAV can accurately detect construction workers with and without helmets in real-time across various site conditions. This proactive approach ensures immediate feedback and intervention, significantly reducing the risk of injuries and fatalities. Additionally, the implementation of UAVs minimizes the workload of site supervisors by automating safety inspections and monitoring, allowing for more efficient and continuous oversight. The experimental results indicate that the UAV system's high precision, recall, and processing capabilities make it a reliable and cost-effective solution for improving construction site safety. The precision, mAP, and FPS of the developed system with the R-CNN are 93.1%, 58.45%, and 27 FPS. This study demonstrates the potential of UAV technology to enhance safety compliance, protect workers, and improve the overall quality of safety management in the construction industry.

Deep Learning-Based Camouflaged Object Detection and Tracking for Enhanced Video Surveillance

To improve security and tactical awareness, video surveillance systems need to be able to find and track objects that are hidden. The approach suggests a deep learning-based method for finding and following hidden objects. There are two major steps in our method: finding and following. During the detecting step, we use a R-CNN based model to find items that are hidden in each frame. A lot of labelled pictures with different kinds of camouflage patterns and objects are used to train this model. The updated Faster R-CNN has more levels and changes that make it better at finding things that are hidden. To keep track of the objects across frames, we use an LSTM-based tracker. The LSTM tracker uses the object's traits and motion information to keep track of its state and guess where it will be in the next frame. To make our method even more effective, we use pre-trained models like MobileNet and VGG-19 for feature extraction. This makes the computations simpler while keeping the accuracy of the recognition and tracking high. The proposed model tested on a number of difficult video scenes with items that were well hidden. The results show that our method works better than other methods, as it increases the accuracy of both recognition and tracking. Also, our method works well even when the lighting changes, the size of the objects changes, or the background is complicated. The suggested method based on deep learning looks like a good way to find and follow hidden objects in video monitoring scenarios. Using CNNs, LSTM networks, and pre-trained models like MobileNet and VGG-19, we show big improvements in terms to compare the various performance parameters with accuracy and stability, which helps make security and monitoring systems better.

Improved Faster R-CNN Based on Multi-Scale Module for Detecting Weak Ground Targets in Remote Sensing Images

In order to optimize the detection capability of small and weak ground targets in remote sensing images, a remote sensing image ground weak target detection method based on multi-scale module improved Faster R-CNN is proposed. By adjusting the Region Proposal Network (RPN) and Region of Interest (RoI) pooling layers, the model is better suited for feature extraction and classification of small targets. Generate feature maps at multiple scales by introducing additional convolutional layers to process inputs at different resolutions, with each layer focusing on targets within a specific size range. These multi-scale features are integrated and input into RPN to provide richer target proposal characteristics. By using image transformation techniques such as scaling, cropping, and rotation, the training dataset is expanded to simulate different small target scenes, enhancing the model’s adaptability to changes in target size, direction, and environment. The experimental results show that the proposed method not only outperforms other models in accuracy, recall, and F1 score, but also maintains high consistency and stability across different datasets. Its significant advantage in mean accuracy (MAP) further proves its reliability and effectiveness in remote sensing applications.

Intelligent Vision System with Pruning and Web Interface for Real-Time Defect Detection on African Plum Surfaces

Agriculture stands as the cornerstone of Africa’s economy, supporting over 60% of the continent’s labor force. Despite its significance, the quality assessment of agricultural products remains a challenging task, particularly at a large scale, consuming valuable time and resources. The African plum is an agricultural fruit that is widely consumed across West and Central Africa but remains underrepresented in AI research. In this paper, we collected a dataset of 2892 African plum samples from fields in Cameroon representing the first dataset of its kind for training AI models. The dataset contains images of plums annotated with quality grades. We then trained and evaluated various state-of-the-art object detection and image classification models, including YOLOv5, YOLOv8, YOLOv9, Fast R-CNN, Mask R-CNN, VGG-16, DenseNet-121, MobileNet, and ResNet, on this African plum dataset. Our experimentation resulted in mean average precision scores ranging from 88.2% to 89.9% and accuracies between 86% and 91% for the object detection models and the classification models, respectively. We then performed model pruning to reduce model sizes while preserving performance, achieving up to 93.6% mean average precision and 99.09% accuracy after pruning YOLOv5, YOLOv8 and ResNet by 10–30%. We deployed the high-performing YOLOv8 system in a web application, offering an accessible AI-based quality assessment tool tailored for African plums. To the best of our knowledge, this represents the first such solution for assessing this underrepresented fruit, empowering farmers with efficient tools. Our approach integrates agriculture and AI to fill a key gap.