Deep Learning Based Vehicle Detection From Aerial Images

This research Paper proposes an application of Image Processing that works on the principle of machine learning and YOLO (You Only Look Once) algorithm was used to detect the custom object. The proposed application is a method to detect non-biodegradable waste from the bins so that the non-biodegradable waste can easily be separated from the bins. Data was divided into three major categories of non- biodegradable waste that were glass, metal and plastic. 450 -500 images of each category were collected in order to train the model. Each image was labelled using a labelling tool in the format of yolo. A machine learning model was created that was trained on the data set of all three categories of the images of non- biodegradable waste. After completion of training a number of files were obtained that were further used in testing model of the application. After completion of testing and checking the accuracy of the model on the webcam all the files were finally imported on the raspberry pi camera. Final testing was done on raspberry pi camera and results were verified accordingly and accuracy was calculated on the basis of the output received.

With the continuous advancement of industrial automation, product quality inspection has become increasingly important in the manufacturing process. Traditional inspection methods, which often rely on manual checks or simple machine vision techniques, suffer from low efficiency and insufficient accuracy. In recent years, deep learning technology, especially the YOLO (You Only Look Once) algorithm, has emerged as a prominent solution in the field of product defect detection due to its efficient real-time detection capabilities and excellent classification performance. This study aims to use the YOLO algorithm to detect and classify defects in product images. By constructing and training a YOLO model, we conducted experiments on multiple industrial product datasets. The results demonstrate that this method can achieve real-time detection while maintaining high detection accuracy, significantly improving the efficiency and accuracy of product quality inspection. This paper further analyzes the advantages and limitations of the YOLO algorithm in practical applications and explores future research directions.

Dental caries, a widespread oral health issue, necessitates early and accurate detection to ensure effective treatment and prevention. Traditional diagnostic methods often involve manual inspection and X-ray imaging, which can be time-consuming and subject to human error. This study introduces a cutting-edge approach by integrating machine learning with the YOLO (You Only Look Once) algorithm and leveraging the computational power of a Raspberry Pi to automate and enhance caries detection. The research focuses on developing a real-time, portable system for detecting dental caries from intraoral images using a Raspberry Pi. The YOLO algorithm, known for its efficiency in real-time object detection, is employed to train a model capable of identifying and classifying carious lesions in dental radiographs with high precision. This study aims to advance dental diagnostic capabilities by combining machine learning and embedded systems technology. The use of the YOLO algorithm with Raspberry Pi not only enhances the accuracy and speed of caries detection but also makes this advanced technology accessible and practical for everyday clinical use. Key Words: Dental caries, Open cv ,raspberry pi4 ,yoloV7

To ensure the safe operation of highway traffic lines, given the imperfect feature extraction of existing road pit defect detection models and the practicability of detection equipment, this paper proposes a lightweight target detection algorithm with enhanced feature extraction based on the YOLO (You Only Look Once) algorithm. The BIFPN (Bidirectional Feature Pyramid Network) network structure is used for multi-scale feature fusion to enhance the feature extraction ability, and Varifocal Loss is used to optimize the sample imbalance problem, which improves the accuracy of road defect target detection. In the evaluation test of the model in the constructed PCD1 (Pavement Check Dataset) dataset, the mAP@.5 (mean Average Precision when IoU = 0.5) of the BV-YOLOv5S (BiFPN Varifocal Loss-YOLOv5S) model increased by 4.1%, 3%, and 0.9%, respectively, compared with the YOLOv3-tiny, YOLOv5S, and B-YOLOv5S (BiFPN-YOLOv5S; BV-YOLOv5S does not use the Improved Focal Loss function) models. Through the analysis and comparison of experimental results, it is proved that the proposed BV-YOLOv5S network model performs better and is more reliable in the detection of pavement defects and can meet the needs of road safety detection projects with high real-time and flexibility requirements.

- The integration of artificial intelligence (AI) into skincare and cosmetics has opened new opportunities for personalized beauty solutions.Introduces a chatbot system that provides tailored cosmetic recommendations based on individual skin types and conditions.Utilizing the YOLO (You Only Look Once) algorithm, the system performs real-time skin analysis by detecting skin types such as acne-prone-skin, burned-skin, dry-skin and oily-skin from user-provided images.The chatbot combines these insights with user input on preferences and concerns to deliver customized product suggestions and skincare routines.The YOLO algorithm's efficiency in object detection ensures rapid and accurate analysis, while the chatbot interface enhances user accessibility and interaction. Key Words: Facial recognition,Skin analysis,Cosmetic suggestions,Chatbot

There are new regulations requiring the use of masks or face shields to prevent the transmission of Covid-19. Using deep learning, a model can be made to detect faces that use masks and face shields by training the model using the previous pre-trained model and using a custom dataset. The purpose of this study is to create a deep learning model that can detect faces with and without masks and as well as face shields for the prevention of covid-19 transmission using YOLO (You Only Look Once) with pre-trained models and custom datasets in real-time. In this study, using pre-trained models from YOLOv3, YOLOv3-Tiny, YOLOv4, YOLOv4-Tiny, and YOLOv4-Tiny-3l with Darknet Framework and compare between average pooling and max pooling in the convolutional neural network YOLO to detect face masks and face shields as a real-time. From experiment the highest mAP was obtained from YOLOv4 using average pooling with a value is 97.64% although the difference is not too much with YOLOv4 using max pooling with value 97.57% and the lowest was YOLOv3-Tiny using max pooling, which was 94.09%, and for the highest FPS was obtained by YOLOv4-Tiny with Fps values is 171 and mAP 96.75%. And for real-time detection of face masks and face shields, the best model used in testing using webcam 1080p is from YOLOv4-Tiny, because the FPS is quite good and the mAP is quite high.

Human wildlife conflict occurs when encounters between humans and wildlife result in negative outcomes such as loss of property, livelihood, and even life for both humans and wildlife, which has a significant impact on the eco system. Interactions between people and wild animals are becoming more common as the human population grows and biodiversity increases. Detecting wild animals and preventing them from entering human habitats, resulting in human-wildlife conflict. It is intended to use machine learning to identify various types of endangered animals. The Dataset contains over 10,000 images in which 7,000 images are training images and 3,000 images are testing images. Our proposed solution employs You only Look once (YOLO) algorithm which is a Convolution Neural Network (CNN) to detect and identify the wild animals using Machine Learning model. The proposed method yields 94% accuracy in detecting the endangered wild animals through the computer vision algorithms. The achieved accuracy percentage is comparatively the highest among the existing model [2] because of the YOLO V4 model.

Advancements in digital imaging and machine learning have transformed healthcare, enabling innovative solutions for automated drug identification. This study develops an image-based system to classify pharmaceutical drugs, tackling errors arising from visual similarities in their shape, color, or size. Accurate drug identification is crucial for healthcare professionals and patients to access reliable information on drug composition, usage instructions, and potential side effects, enhancing safety and efficiency in medical practice. The system leverages the YOLO (You Only Look Once) algorithm, renowned for its speed and precision in object detection. A dataset comprising 5,000 drug images sourced from Kaggle was curated, with annotations and augmentation techniques such as horizontal flipping, rotation, and scaling to improve model robustness. The YOLOv11 model achieved a precision of 97.4%, a recall of 97.6%, and a mean average precision (mAP@50) of 98.4%, demonstrating high reliability in real-world scenarios. Integrated with a user-friendly Tkinter interface, the system facilitates real-time drug detection and information retrieval, streamlining access to critical data. This research underscores the YOLO algorithm’s effectiveness in delivering rapid and accurate drug classification, offering a scalable solution for healthcare applications. The system’s success highlights its potential to reduce medication errors and improve patient outcomes through precise and accessible drug identification technology.

This research aimed to develop detection model for aflatoxin-contaminated cocoa beans using YOLO (You Only Look Once) method based on fluorescence imaging. Aflatoxin level measurement and image acquisition were carried out on cocoa beans inoculated with Aspergillus flavus (6 mL/100 g) and incubated for 7 days. The quantification of aflatoxin levels was achieved through Liquid Mass Chromatography (LCMS), which served as the basis for categorizing the images into 3 groups, namely aflatoxin-free, contaminated below and contaminated above the threshold. The image acquisition was conducted in a mini studio equipped with UV light. Furthermore, image annotation was carried out using Roboflow, while YOLOv5 was employed as detection model for aflatoxin-contaminated cocoa beans. The performance of YOLO model, based on testing data, yielded the following metrics: precision (all): 0.91, recall (all): 0.93, mean average precision (all): 0.95 and mAP@[.5:.95] (all): 0.47. The average accuracy based on the confusion matrix from the testing data was 91%. These results showed that a combination of YOLOv5 model and fluorescence imaging presented a highly performing alternative for detecting various aflatoxin contamination levels in cocoa beans.

Cardiovascular diseases (CVDs) remain a leading cause of global mortality, underscoring the urgent need for early and accurate diagnosis. Electrocardiograms (ECGs) are a widely used, noninvasive tool for detecting cardiac abnormalities, but manual interpretation can be time- consuming and prone to error. This study explores advanced machine learning (ML) and deep learning (DL) methods to improve the detection of CVDs from ECG images, leveraging their ability to analyze complex patterns with high precision. Models such as Convolutional Neural Networks (CNNs), Vision Transformers (ViTs), and ensemble frameworks are utilized, alongside traditional ML models like XGBoost, CatBoost, and LightGBM. The performance of these models is evaluated based on precision, recall, F1-score, and accuracy. Results show that DL models, particularly an ensemble of CNN and ViT, achieve superior accuracy , outperforming standalone ML methods. Furthermore, an ensemble of XGBoost, CatBoost, and LightGBM achieves competitive results, while Optuna-tuned XGBoost further enhances ML performance. The findings highlight the potential of combining ML and DL techniques for robust and reliable CVD detection in ECG images. This research demonstrates the promise of AI-driven solutions in advancing medical diagnostics and supports the integration of automated systems for early detection of cardiovascular diseases.

Real-time deep learning-based position control of a mobile robot

Automatic inspection of insulators from high-voltage transmission lines is of paramount importance to the safety and reliable operation of the power grid. Due to different size insulators and the complex background of aerial images, it is a difficult task to recognize insulators in aerial views. Most of the traditional image processing methods and machine learning methods cannot achieve sufficient performance for insulator detection when diverse background interference is present. In this study, a deep learning method—based on You Only Look Once (YOLO)—will be proposed, capable of detecting insulators from aerial images with complex backgrounds. Firstly, aerial images with common aerial scenes were collected by Unmanned Aerial Vehicle (UAV), and a novel insulator dataset was constructed. Secondly, to enhance feature reuse and propagation, on the basis of YOLOv3 and Dense-Blocks, the YOLOv3-dense network was utilized for insulator detection. To improve detection accuracy for different sized insulators, a structure of multiscale feature fusion was adapted to the YOLOv3-dense network. To obtain abundant semantic information of upper and lower layers, multilevel feature mapping modules were employed across the YOLOv3-dense network. Finally, the YOLOv3-dense network and compared networks were trained and tested on the testing set. The average precision of YOLOv3-dense, YOLOv3, and YOLOv2 were 94.47%, 90.31%, and 83.43%, respectively. Experimental results and analysis validate the claim that the proposed YOLOv3-dense network achieves good performance in the detection of different size insulators amid diverse background interference.

Deep learning (DL) and machine learning (ML) methods have recently contributed to the advancement of models in the various aspects of prediction, planning, and uncertainty analysis of smart cities and urban development. This paper presents the state of the art of DL and ML methods used in this realm. Through a novel taxonomy, the advances in model development and new application domains in urban sustainability and smart cities are presented. Findings reveal that five DL and ML methods have been most applied to address the different aspects of smart cities. These are artificial neural networks; support vector machines; decision trees; ensembles, Bayesians, hybrids, and neuro-fuzzy; and deep learning. It is also disclosed that energy, health, and urban transport are the main domains of smart cities that DL and ML methods contributed in to address their problems.

Automatic target recognition (ATR) is a challenging task for several computer vision applications. It requires efficient, accurate, and robust methods for target detection and target identification. Deep learning has shown great success in many computer vision applications involving color RGB images. However, the performance of these networks in ATR with infrared sensor data needs further investigation. In this paper, we propose a multistage automatic target detection and recognition (ATDR) system that performs both target detection and target classification on infrared (IR) imagery using deep learning. Our system processes large IR image frames where targets take <1 % of the total number of pixels. First, we train a state-of-the-art object detector you only look once (YOLO) to localize all potential targets in the input image frame. Then, we train a convolutional neural network (CNN) to identify these detections as targets or false alarms. In this second phase, we adapt and analyze the performance of three CNN architectures: a compact and fully connected CNN, VGG16 with batch normalization, and a wide residual neural network (WRN). We also explore the use of a loss function that optimizes directly the area under the receiver operating characteristic (ROC) curve (AUC), and adapt it to our ATR application. To enhance the robustness of the proposed ATR to perturbation and variations introduced during the detection stage, we train our CNN classifiers on automatically detected targets using YOLO, in addition to ground truth bounding boxes and apply selected data augmentation techniques. To simulate real testing environments, where the spatial location of the targets within the image frame is unknown, only YOLO-detected boxes are used during validation. We evaluate our ATDR on a real benchmark dataset that includes different vehicles captured at different resolutions. Our experiments have shown that YOLO can detect most of the targets at the expense of generating a high number of false alarms. We show that the VGG-16 network with batch normalization, which is the best performing model, can correctly identify the classes of the targets, as well as classify the majority of YOLO’s false detections into an additional nontarget class. We also show that the proposed training modification to optimize an AUC-based loss function for ATR proved to be advantageous mainly in identifying difficult targets.

Among the population of India, nearly 1.6 million children are blind or visually impaired by birth. The challenges faced by the visually impaired persons are high in day-to-day life. There are number of challenges faced by them like requiring guidance for the usage of public transport, to walk independently. One of the most common problems is the dependency on others for the purchase of household items. Hence, a unique solution is required which makes the people to identify and purchase the products in the supermarket without dependency. This research focuses on a portable camera-based assistance that helps them to identify the grocery items along with the framework to read the text on the labels. This experimental system consists of three modules: object detection and classification; reading the text from the label; text to audio converter. The input to the camera is fed by the Raspberry Pi kit. The captured video is segregated as pictures of the grocery items. The proposed deep learning algorithm for the detection of object is You only look once (YOLO). The accuracy of the detection is pre-dominantly high compared to the existing algorithms. The captured image of the grocery item label is preprocessed to blur the unwanted text. Optical character recognition (OCR) is applied over the label to convert as the text where machine can read. At last, the processed text is converted as audio for the guidance of the people. Thus, a flexible supervisory mechanism with user-friendliness is developed for helping the visually impaired people.KeywordsDeep learningVisually impairedSmart super market assistanceOptical character recognition