Deep Learning Object Detection Research Articles

Design and Implementation for BIC Code Recognition System of Containers using OCR and CRAFT in Smart Logistics

The BIC (Bureau International des Containers et du Transport Intermodal) Code is the identification code for ocean shipping containers and is crucial for logistics, transportation, and security. Accurate recognition of container BIC Code is essential for efficient import and export processes, authorities' ability to intercept illegal goods and safe transportation. Nevertheless, the current practice of employees recognizing and manually entering container BIC codes is inefficient and prone to error. Although automated recognition efforts have been made, challenges remain due to the aging of containers, manufacturing differences between companies, and the mixing of letters and numbers in the 11-digit combination. In this paper, we propose the design and implementation of a BIC Code recognition system using an open source-based OCR engine, deep learning object detection algorithm, and text detector model. In the logistics industry, various attempts are being made to seamlessly link the data required at each stage of transportation between these systems. If we can secure the stability and consistency of BIC Code recognition that can be used in the field through our research, it will contribute to overcoming the instability caused by false positives.

Development and deployment of a big data pipeline for field-based high-throughput cotton phenotyping data

In this study, we propose a big data pipeline for cotton bloom detection using a Lambda architecture, which enables real-time and batch processing of data. Our proposed approach leverages Azure resources such as Data Factory, Event Grids, Rest APIs, and Databricks. This work is the first to develop and demonstrate the implementation of such a pipeline for plant phenotyping through Azure's cloud computing service. The proposed pipeline consists of data preprocessing, object detection using a YOLOv5 neural network model trained through Azure AutoML, and visualization of object detection bounding boxes on output images. The trained model achieves a mean Average Precision (mAP) score of 0.96, demonstrating its high performance for cotton bloom classification. We evaluate our Lambda architecture pipeline using 9,000 images yielding an optimized runtime of 34 minutes. The results illustrate the scalability of the proposed pipeline as a solution for deep learning object detection, with the potential for further expansion through additional Azure processing cores. This work advances the scientific research field by providing a new method for cotton bloom detection on a large dataset and demonstrates the potential of utilizing cloud computing resources, specifically Azure, for efficient and accurate big data processing in precision agriculture.

Vehicle-camel collisions in Saudi Arabia: Application of single and multi-stage deep learning object detectors

Vehicle-camel collision is a persistent issue in countries where population of camels is high such as Saudi Arabia. The purpose of the research is to introduce a new solution to eliminate this issue. Previous solutions, such as fencing the sides of the roads, designing better camel warning signs and fining camel owners when camels cross high traffic roads, are either expensive, ineffective, or hard to implement. Therefore, in this work, we harness the power of deep learning to tackle this problem. In particular, we use state-of-the-art deep learning object detectors to detect camels on roads with high accuracy. Results show that all implemented models were capable of detecting camels on or near roads. Moreover, the single-stage detector Yolo v3 was found to be the most accurate and is as fast as its successor Yolo v4. Findings of this work helped select the deep learning model needed for a reliable and automatic vehicle-camel collision avoidance system.

WT-YOLOM: An Improved Target Detection Model Based on YOLOv4 for Endogenous Impurity in Walnuts

Since impurities produced during walnut processing can cause serious harm to human health, strict quality control must be carried out during production. However, most detection equipment still uses photoelectric detection technology to automatically sort heterochromatic particles, which is unsuitable for detecting endogenous foreign bodies with similar colors. Therefore, this paper proposes an improved YOLOv4 deep learning object detection algorithm, WT-YOLOM, for detecting endogenous impurities in walnuts—namely, oily kernels, black spot kernels, withered kernels, and ground nutshells. In the backbone of the model, a lightweight MobileNet module was used as the encoder for the extraction of features. The spatial pyramid pooling (SPP) structure was improved to spatial pyramid pooling—fast (SPPF), and the model size was further reduced. Loss function was replaced in this model with a more comprehensive SIoU loss. In addition, efficient channel attention (ECA) mechanisms were applied after the backbone feature map to improve the model’s recognition accuracy. This paper compares the recognition speed and accuracy of the WT-YOLOM algorithm with the Faster R-CNN, EfficientDet, CenterNet, and YOLOv4 algorithms. The results showed that the average precision of this model for different kinds of endogenous impurities in walnuts reached 94.4%. Compared with the original model, the size was reduced by 88.6%, and the recognition speed reached 60.1 FPS, which was an increase of 29.0%. The metrics of the WT-YOLOM model were significantly better than those of comparative models and can significantly improve the detection efficiency of endogenous foreign bodies in walnuts.

Food Classification and Meal Intake Amount Estimation through Deep Learning

This paper proposes a method to classify food types and to estimate meal intake amounts in pre- and post-meal images through a deep learning object detection network. The food types and the food regions are detected through Mask R-CNN. In order to make both pre- and post-meal images to a same capturing environment, the post-meal image is corrected through a homography transformation based on the meal plate regions in both images. The 3D shape of the food is determined as one of a spherical cap, a cone, and a cuboid depending on the food type. The meal intake amount is estimated as food volume differences between the pre-meal and post-meal images. As results of the simulation, the food classification accuracy and the food region detection accuracy are up to 97.57% and 93.6%, respectively.

Application access for chromatic color detection

The process of identifying specific examples of an item inside an image or video is referred to as “object detection,” and it is one of the computer vision techniques. When individuals look at pictures or movies, they are able to recognize and find things that are of interest to them. The primary goal of the field of object recognition is to achieve this level of intelligence via the use of a computer. Image segmentation and blob analysis, which uses simple object properties such as size, shape, or color, and Feature-based object detection, which uses feature extraction, matching, and RANSAC to estimate the location of an object, are two of the many common techniques that are used in conjunction with deep learning and machine learning-based object detection. There are also many other common techniques that are used in conjunction with deep learning and machine learning-based object detection. The most challenging aspects of object detection are the processes of categorizing things and pinpointing their locations. When seen from a different perspective, the same object might seem to be completely different. Detection algorithms have a tough time distinguishing between various things at different scales and perspectives since the sizes and proportions of items almost always alter. The camera input would be used to determine an object’s shape, according to this project’s proposal. Circle, square, and triangle are the three distinct shapes that we are determining an item to be right now. The finished result is shown in MATLAB’s graphical user interface (GUI). The productive identification process is carried out with the assistance of MATLAB. Using the SFTA Algorithm, the application extracts features from the specific image of a video. The neural network algorithm is used to classify the image of video. Image segmentation and image detection are two examples of related computer vision methods that are inextricably intertwined with object recognition. These techniques allow us to better comprehend and analyze the situations shown in movies and still photographs. It has been determined whether or not the suggested system is suitable. Because of this, the testing has shown that the suggested system generates generally accurate indications of real performance at a rate of 99.4%.

A Review on Real Time Object Detection Using Deep Learning

Abstract: In object detection system we implement various convolutional networks to acquire a precise recognition and localization of objects in digital images and videos. This system sees real time applications in security, diagnosis, surveillance, automation, sports and many others. YOLO is one of the available models of deep learning which is used for quick implementation of object detection system. We will have a brief study of deep learning for object detection in this paper

U-YOLOv7: A network for underwater organism detection

Detecting and monitoring underwater organisms is very important for sea aquaculture. The human eye struggles to quickly distinguish between aquatic species due to their variety and dense dispersion. In this paper, a deep learning object detection algorithm based on YOLOv7 is used to design a new network, called Underwater-YOLOv7 (U-YOLOv7), for underwater organism detection. This model satisfies the requirements with regards to both speed and accuracy. First, a network combining CrossConv and an efficient squeeze-excitation module is created. This network increases the extraction of channel information while reducing parameters and enhancing the feature fusion of the network. Second, a lightweight Content-Aware ReAssembly of FEatures (CARAFE) operator is used to obtain more semantic information about underwater images before feature fusion. A 3D attention mechanism is incorporated to improve the anti-interference ability of the model in underwater recognition. Finally, a decoupling head using hybrid convolution is designed to accelerate convergence and improve the accuracy of underwater detection. The results show that the network proposed in this paper obtains an improvement of 3.2% in accuracy, 2.3% in recall, and 2.8% in the mean average precision value and runs at up to 179 fps, far outperforming other advanced networks. Moreover, it has a higher estimation accuracy than the YOLOv7 network.

Logistics box recognition in robotic industrial de-palletising procedure with systematic RGB-D image processing supported by multiple deep learning methods

In an automated box de-palletisation system that utilises robots, vision-based box recognition on the pallet plays the main role in providing picking guidelines. The complexity of the working condition and the target object, particularly the cluttered arrangement and various outer surfaces of the boxes, significantly affect the quality of the outcome. Typically, a large-scale vision dataset is required to train a deep learning object-detection model. However, considerable effort and time is required to achieve this. Therefore, this study proposes a Mask R-CNN-based detection approach for box objects, which is supported by a cycle generative adversarial network (Cycle GAN). The purpose of the Cycle-GAN is to optimise the outer surfaces of boxes by automatically erasing tags, stickers, labels, and symbols that exist on the boxes before loading them to the Mask R-CNN for detection. Subsequently, the obtained result was combined with the output from the developed boundary-enhancing technique that was applied to a depth map. Consequently, the box detection performance was significantly improved, and it was confirmed through experiments with a practical robot system in picking tasks. In the experiments, the success rate of the proposed method was validated using 200 cases of orderly and disorderly arrangements of boxes, respectively. Furthermore, the metric of the mean absolute error between the predicted picking point and the ground truth values for the test cases in the implementation process for the robot operation was also researched.

An Ensemble of Deep Learning Object Detection Models for Anatomical and Pathological Regions in Brain MRI.

This paper proposes ensemble strategies for the deep learning object detection models carried out by combining the variants of a model and different models to enhance the anatomical and pathological object detection performance in brain MRI. In this study, with the help of the novel Gazi Brains 2020 dataset, five different anatomical parts and one pathological part that can be observed in brain MRI were identified, such as the region of interest, eye, optic nerves, lateral ventricles, third ventricle, and a whole tumor. Firstly, comprehensive benchmarking of the nine state-of-the-art object detection models was carried out to determine the capabilities of the models in detecting the anatomical and pathological parts. Then, four different ensemble strategies for nine object detectors were applied to boost the detection performance using the bounding box fusion technique. The ensemble of individual model variants increased the anatomical and pathological object detection performance by up to 10% in terms of the mean average precision (mAP). In addition, considering the class-based average precision (AP) value of the anatomical parts, an up to 18% AP improvement was achieved. Similarly, the ensemble strategy of the best different models outperformed the best individual model by 3.3% mAP. Additionally, while an up to 7% better FAUC, which is the area under the TPR vs. FPPI curve, was achieved on the Gazi Brains 2020 dataset, a 2% better FAUC score was obtained on the BraTS 2020 dataset. The proposed ensemble strategies were found to be much more efficient in finding the anatomical and pathological parts with a small number of anatomic objects, such as the optic nerve and third ventricle, and producing higher TPR values, especially at low FPPI values, compared to the best individual methods.

Small Pests Detection in Field Crops Using Deep Learning Object Detection

Deep learning algorithms, such as convolutional neural networks (CNNs), have been widely studied and applied in various fields including agriculture. Agriculture is the most important source of food and income in human life. In most countries, the backbone of the economy is based on agriculture. Pests are one of the major challenges in crop production worldwide. To reduce the overall production and economic loss from pests, advancement in computer vision and artificial intelligence may lead to early and small pest detection with greater accuracy and speed. In this paper, an approach for early pest detection using deep learning and convolutional neural networks has been presented. Object detection is applied on a dataset with images of thistle caterpillars, red beetles, and citrus psylla. The input dataset contains 9875 images of all the pests under different illumination conditions. State-of-the-art Yolo v3, Yolov3-Tiny, Yolov4, Yolov4-Tiny, Yolov6, and Yolov8 have been adopted in this study for detection. All of these models were selected based on their performance in object detection. The images were annotated in the Yolo format. Yolov8 achieved the highest mAP of 84.7% with an average loss of 0.7939, which is better than the results reported in other works when compared to small pest detection. The Yolov8 model was further integrated in an Android application for real time pest detection. This paper contributes the implementation of novel deep learning models, analytical methodology, and a workflow to detect pests in crops for effective pest management.

Research on Video Synopsis Based on Deep Learning Target Detection and Target Trajectory Optimization

In this era of rapid economic and technological development, monitoring is essential for security issues. However, it is also accompanied by the low efficiency of managing and viewing a large number of surveillance videos every day. Traditional methods are gradually unable to meet the requirements of application, and the video field requires algorithm technology that can better solve practical problems. Intelligent monitoring technology is gradually emerging and developing, based on theories such as computer vision, which can solve many monitoring video problems and greatly improve the efficiency of monitoring video work. Through in-depth exploration of existing video synopsis technology algorithms, this article proposes a new video synopsis method that can effectively detect and track videos, thereby greatly improving the efficiency of storage, transmission, and use of surveillance videos. Experiments have shown that this method can effectively concentrate surveillance videos, and compared with existing methods, it has a better synopsis ratio while ensuring the integrity of video information, effectively reducing collisions between targets, effectively reducing overlap between targets, and achieving good visual effects. Utilizing an improved deep learning object detection and multi object tracking algorithm with added attention mechanism to extract foreground moving targets in videos, and using a mixed Gaussian background modeling algorithm establish a background, laying a stable target foundation for subsequent video concentration. At the same time, design trajectory recombination optimization methods to ensure that the targets do not overlap as much as possible. Reasonably place the targets in the new condensed video sequence, determine the index of all targets in the new condensed video stream, and finally integrate each frame of target images into the background image according to the set index rules, ultimately obtaining the synopsis video.

An Edge-computing flow meter reading recognition algorithm optimized for agricultural IoT network

Groundwater resources in Nebraska, U.S. are closely monitored by 23 Natural Resources Districts (NRDs) located across the state. Growers who use groundwater for irrigation are required to have flow meters installed at wells to monitor their water usage. However, many of these flow meters are still being read and recorded through in-person visits, which can be time-consuming and costly. Although some flow meters in Nebraska are monitored remotely by telemetry-enabled camera systems, yearly telemetry costs are high and making long-term operation financially burdensome. Using less expensive network protocol, such as Internet of Things (IoT), to transmit flow meter readings could enable new monitoring opportunities. However, there are challenges in directly transmitting flow meter images via IoT due to limited bandwidth. Therefore, in this study, we developed an algorithm using object detection deep learning techniques, i.e. You Only Look Once (YOLO) that can be programmed at an IoT node which can recognize readings from images of flow meters onsite before transmitting. The developed algorithm could significantly reduce data size and is essential for flow meter monitoring in an IoT network setting. The developed algorithm achieved 95.35% accuracy when recognizing 1,248 real-world flow meter images obtained at the courtesy of North Platte Natural Resources District (NPNRD) in western Nebraska. The framework and algorithm were also tested in a real-world scenario on a flow meter installed on a linear-move sprinkler irrigation system and showed promising results. By leveraging IoT and deep learning techniques, this research has the potential to revolutionize flow meter monitoring, reducing costs and improving efficiency in the management of groundwater resources in Nebraska, and potentially in other regions as well.

Deep learning in detecting tumor spread through air spaces

Lung cancer is a leading cause of death on a global scale. Professor I-Fang Chung leads a team of researchers at the Institute of Biomedical Informatics, National Yang Ming Chiao Tung University, Taiwan, investigating how machine learning can help predict the risk of tumour recurrence following lung cancer surgery. The standard treatment for early-stage lung cancer patients is complete surgical resection of the tumour but disease recurrence within the first five years following surgery is common. The route of this recurrence is usually tumour spread through the air spaces (STAS) and the presence of STAS has recently been identified as a risk factor for recurrence of tumours. The researchers are exploring the importance of STAS and how targeting this phenomenon and using machine learning methods to aid in the analysis of medical imaging of diseased tissues can help in predicting the recurrence risk of post-surgery lung cancer patients. Dr Yi-Chen Yeh, a pathologist from the Taipei Veterans General Hospital, is working with Chung to employ a variety of deep-learning object detection methodologies to detect STAS in pathology images. Regions of interest (ROI) images extracted from pathology whole slide images (WSI) are marked and annotated, which provides location information for STAS. Deep learning object detection methods are used to train a model which can find STAS, then additional techniques including using pre-trained model parameters, augmenting random image data and modifying loss function are used to improve the detection rates for the model. Their model helps pathologists identify STAS and accurately predicts patient outcomes.

Automatic Mapping of Physical Urban Problems Using Remotely Sensed Imagery

While big cities are expected to exercise cost-effective, evidence-based planning, many are under reactive management, facing simultaneous problems and limited resources. This project develops a proof-of-concept workflow for the automatic monitoring of physical urban problems by combining remote sensing for detection and cartography for visualization. The example problem treated was the obstructive parking of vehicles on pavements as proxy for restricted urban mobility. Nine aerial images of UK urban areas were processed by a deep learning object detector of standard cars, achieving an F-score of 70.72%. Two large scale map reports of 200m wide areas were produced, featuring car detections and overlaps with topographic mapping features. Complementary analysis included the calculation of total detection window overlap per roadside pavement and its change with time. The proposed method combines uniform city-wide coverage with fast interpretation and can inspire the development of professional urban planning tools.

Application of YOLOv5 for Point Label Based Object Detection of Black Pine Trees with Vitality Losses in UAV Data

Monitoring tree diseases in forests is crucial for managing pathogens, particularly as climate change and globalization lead to the emergence and spread of tree diseases. Object detection algorithms for monitoring tree diseases through remote sensing rely on bounding boxes to represent trees. However, this approach may not be the most efficient. Our study proposed a solution to this challenge by applying object detection to unmanned aerial vehicle (UAV)-based imagery, using point labels that were converted into equally sized square bounding boxes. This allowed for effective and extensive monitoring of black pine (Pinus nigra L.) trees with vitality-related damages. To achieve this, we used the “You Only Look Once’’ version 5 (YOLOv5) deep learning algorithm for object detection, alongside a 16 by 16 intersection over union (IOU) and confidence threshold grid search, and five-fold cross-validation. Our dataset used for training and evaluating the YOLOv5 models consisted of 179 images, containing a total of 2374 labeled trees. Our experiments revealed that, for achieving the best results, the constant bounding box size should cover at least the center half of the tree canopy. Moreover, we found that YOLOv5s was the optimal model architecture. Our final model achieved competitive results for detecting damaged black pines, with a 95% confidence interval of the F1 score of 67–77%. These results can possibly be improved by incorporating more data, which is less effort-intensive due to the use of point labels. Additionally, there is potential for advancements in the method of converting points to bounding boxes by utilizing more sophisticated algorithms, providing an opportunity for further research. Overall, this study presents an efficient method for monitoring forest health at the single tree level, using point labels on UAV-based imagery with a deep learning object detection algorithm.

A challenge of deep-learning-based object detection for hair follicle dataset.

Deep-learning object detection has been applied in various industries, including healthcare, to address hair loss. In this paper, YOLOv5 object detection algorithm was used to detect hair follicles in a small and specific image dataset collected using a specialized camera on the scalp of individuals with different ages, regions, and genders. The performance of YOLOv5 was compared with other popular object detection models. YOLOv5 performed well in the detection of hair follicles, and the follicles were classified into five classes based on the number of hairs and the type of hair contained. In single-class object detection experiments, a smaller batch size and the smallest YOLOv5s model achieved the best results, with an map of 0.8151. In multiclass object detection experiments, the larger YOLOv5l model was able to achieve the best results, and batch size affected the result of model training. YOLOv5 is a promising algorithm for detecting hair follicles in a small and specific image dataset, and its performance is comparable to other popular object detection models. However, the challenges of small-scale data and sample imbalance need to be addressed to improve the performance of target detection algorithms.

FRCNN-AA-CIF: An automatic detection model of colon polyps based on attention awareness and context information fusion

It is noted that the foreground and background of the polyp images detected under colonoscopy are not highly differentiated, and the feature map extracted by common deep learning object detection models keep getting smaller as the number of networks increases. Therefore, these models tend to ignore the details in pictures, resulting in a high polyp missed detection rate. To reduce the missed detection rate, this paper proposes an automatic detection model of colon polyps based on attention awareness and context information fusion (FRCNN-AA-CIF) based on a two-stage object detection model Faster Region-Convolutional Neural Network (FR-CNN). First, since the addition of attention awareness can make the feature extraction network pay more attention to polyp features, we propose an attention awareness module based on Squeeze-and-Excitation Network (SENet) and Efficient Channel Attention Module (ECA-Net) and add it after each block of the backbone network. Specifically, we first use the 1*1 convolution of ECA-Net to extract local cross-channel information and then use the two fully connected layers of SENet to reduce and increase the dimension, to filter out the channels that are more useful for feature learning. Further, because of the presence of air bubbles, impurities, inflammation, and accumulation of digestive matter around polyps, we used context information around polyps to enhance the focus on polyp features. In particular, after the network extracts the region of interest, we fuse the region of interest with its context information to improve the detection rate of polyps. The proposed model was tested on the colonoscopy dataset provided by Huashan Hospital. Numerical experiments show that FRCNN-AA-CIF has the highest detection accuracy (mAP of 0.817), the lowest missed detection rate of 4.22%, and the best classification effect (AUC of 95.98%). Its mAP increased by 3.3%, MDR decreased by 1.97%, and AUC increased by 1.8%. Compared with other object detection models, FRCNN-AA-CIF has significantly improved recognition accuracy and reduced missed detection rate.

Spot-the-Camel: Computer Vision for Safer Roads

As the population grows and more land is being used for urbanization, ecosystems are disrupted by our roads and cars. This expansion of infrastructure cuts through wildlife territories, leading to many instances of Wildlife-Vehicle Collision (WVC). These instances of WVC are a global issue that is having a global socio-economic impact, resulting in billions of dollars in property damage and, at times, fatalitiesfor vehicle occupants. In Saudi Arabia, this issue is similar, with instances of Camel-Vehicle Collision (CVC) being particularly deadly due to the large size of camels, which results in a 25% fatality rate [1].The focus of this work is to test different object detection models on the task of detecting camels on theroad. The Deep Learning (DL) object detection models used in the experiments are: CenterNet, Efficient Det, Faster R-CNN, SSD, and YOLOv8. Results of the experiments show that YOLOv8 performed the best in terms of accuracy and was the most efficient in training. In the future, the plan is to expand on this work by developing a system to make countryside roads safer.

Banana Pseudostem Visual Detection Method Based on Improved YOLOV7 Detection Algorithm

Detecting banana pseudostems is an indispensable part of the intelligent management of banana cultivation, which can be used in settings such as counting banana pseudostems and smart fertilization. In complex environments, dense and occlusion banana pseudostems pose a significant challenge for detection. This paper proposes an improved YOLOV7 deep learning object detection algorithm, YOLOV7-FM, for detecting banana pseudostems with different growth conditions. In the loss optimization part of the YOLOV7 model, Focal loss is introduced, to optimize the problematic training for banana pseudostems that are dense and sheltered, so as to improve the recognition rate of challenging samples. In the data augmentation part of the YOLOV7 model, the Mixup data augmentation is used, to improve the model’s generalization ability for banana pseudostems with similar features to complex environments. This paper compares the AP (average precision) and inference speed of the YOLOV7-FM algorithm with YOLOX, YOLOV5, YOLOV3, and Faster R-CNN algorithms. The results show that the AP and inference speed of the YOLOV7-FM algorithm is higher than those models that are compared, with an average inference time of 8.0 ms per image containing banana pseudostems and AP of 81.45%. This improved YOLOV7-FM model can achieve fast and accurate detection of banana pseudostems.