JOURNAL OF IMAGE PROCESSING & PATTERN RECOGNITION PROGRESS

Hand Gesture Recognition using Machine Learning and OpenCV

2023-08-03T09:54:04+00:00

People who are deaf or dumb use sign language to communicate with one another and within their own communities. The process of computer recognition of sign language begins with learning sign gestures and progresses until text or speech is generated. Static and dynamic sign gestures are the two categories. Although static gesture recognition is simpler to use than dynamic gesture recognition, both gesture recognition systems are essential to the survival of the human species. The processes for sign language recognition are described in this survey. Data gathering, preprocessing, transformation, feature extraction, classification, and results are all examined. Additionally, there were some suggestions for developing this area of work.

Utilizing ML for Hand Gesture Recognition

2023-08-03T10:43:28+00:00

Hand gestures are an instinctive and common method of communication during human interactions, representing a form of non-verbal expression. Gesture recognition technology aims to interpret and classify meaningful movements performed by human hands. The motivation behind its development is to revolutionize human-computer interaction, addressing drawbacks found in current systems. The study presents a novel algorithm that eliminates the requirement for image background filtering. This algorithm proves versatile, capable of recognizing various hand gestures and accurately determining the count of raised fingers. By focusing on the hand movements within the region of interest, the proposed system enhances the efficiency and precision of gesture recognition. Gesture recognition has extensive applications, spanning from virtual and augmented reality to healthcare and industrial automation. Through this technology, users can engage more naturally and intuitively with digital environments, while healthcare professionals can provide real-time feedback during rehabilitation exercises. In industrial settings, gesture recognition enables the control of machines and robots, enhancing productivity and reducing manual labor. Advancements in computer vision, machine learning (ML), and sensor technology have greatly improved the accuracy and effectiveness of gesture recognition systems. With further research and development, gesture recognition is poised to revolutionize human-computer interactions across diverse fields, enriching user experiences and optimizing task performance. By continuously refining algorithms and incorporating innovative techniques, gesture recognition is poised to become an integral part of future human-machine interactions.

Object’s Area Measurement using Image Processing

2023-10-01T07:27:17+00:00

Object measurement holds substantial significance across a wide spectrum of domains, spanning from quality control in manufacturing to applications in healthcare and environmental monitoring. In this study, a method is suggested to detect the object and calculate the area of the given object in real time using MATLAB. Also, comparing the result obtained with the known area and seeing how accurate the area calculated is. The proposed method for detecting objects and calculating the area is reliable and accurate compared to the area calculated is verified with the manual area calculation readings. A methods is suggested without any constraint of the height from which the object is captured. In this method, the object whose area is to be measured is kept on a white backgroung of known area, the backgroud edges are etracted, the image is thresholded such that the white background has pixel value 255 and the object to be measured has pixel value 0, the image is filled and thus the pixels values initially having 0 intensity now possess intensity level of 255, further, the pixel sum is calculated to find out the calibration factor (area per pixel); lastly, the object is extracted from background and the pixel sum of object is multiplied by the calibration factor to obtain its area. The proposed method was evaluated on the two sets of images: (1) rectangular object; and (2) circular object. The authors respond to the crucial requirement for accurate object measurement by presenting a systematic approach that includes image pre-processing, boundary detection, and calibration factor computation. The proposed methodology is rigorously validated through a series of empirical experiments, revealing its efficacy in accurately quantifying objects with an error of 0.02 cm². This work constitutes a substantial contribution to the field of computer vision, offering a practical and versatile solution for object measurement tasks with broad applications.

CNN-based Approach for Efficient Bell Pepper Leaf Disease Recognition

2023-10-21T06:24:42+00:00

Convolutional Neural Networks (CNNs) have achieved remarkable results in the detection of diseased plant leaves, providing highly accurate predictions. This project offers a comprehensive examination of existing systems designed for the detection of plant-based diseases. Using CNN trained on a dataset of bell pepper plant images, various simulation approaches for neurons and layers were employed. Plant diseases have significant impacts on agricultural productivity, leading to economic losses, reduced crop quality, and decreased yield. Consequently, the timely detection of plant diseases in large crop fields has garnered increased attention. By obtaining reliable data on plant health and accurately identifying diseases, effective management strategies can be implemented to mitigate the spread of diseases. Therefore, our model plays a crucial role in the classification of healthy and diseased bell pepper plant leaves. In this context, it is highly recommended to promptly remove any infected plants to prevent the spread of disease throughout the entire garden. Taking immediate action upon identifying issues with the pepper crop helps minimize further contamination and ensures better overall crop health. (By using Convolutional neural networks (CNN) we obtained maximum accuracy of 99.99%.)

A Brief Survey on Content-based Fake News Detection

2023-08-03T10:06:26+00:00

The increase in spread of fake news causes destruction of democracy and public confidence, which has boosted the demand for reliable fake news identification dramatically. Recent developments in this area have opened up innovative approaches for identifying fake news by looking at how it travels on social media. However, in order to spot false news early, one has no information about news distribution because at an early-stage, fake news can only be created and later dissemination happens. Also, the speed and amount at which news is created and propagated online is destructive in nature. As a result, there is an urgent need to create methods for detecting false news just based on news content and eliminate the danger before it reaches the mass of people. This study is divided into five sections. We have an introduction where we describe the topic and the motivations behind this survey. Fundamental Theories is the second section where we discuss the current theories that are used to detect fake news. In the third section, we briefly touched upon the advantages of utilizing content-based detection. In the fourth section, we conducted a survey on content-based fake news detection and listed the advantages and disadvantages of the available techniques. We closed the study with a discussion of some potential topics to investigate further.