Volume 12 | Issue 5

 Abstract

Agriculture is one of the main factor that decides the economic growth of any country. This is the one of the reasons that disease detection in plants plays an important role in agriculture field, as having disease in plants are quite natural. Plant disease identification is a significant process to prevent the losses in the quality and quantity of the agricultural product. It is essential to detect any disease in time to ensure healthy and proper growth of the plants prior to applying required treatment to the affected plants. Since manual detection of diseases costs a large amount of time and labour, it is inevitably prudent to have an automated system.With the worldwide increase in digital cameras and continuous improvement in computer vision domain, the automated techniques for detection of disease are highly possible. After necessary pre processing, the dataset was trained on using different deep learning algorithms. This approach of ours is to increase the productivity of crops in agriculture.We aim to raise awareness about the disease and also provide solutions to the disease using generative AI.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Agriculture, Plant disease, Deep learning algorithms, Generative AI

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The project's primary aim is to develop an efficient system specialized in recognizing handwritten text, facilitating the smooth conversion of handwritten text images into digital text format. Utilizing cutting-edge machine learning techniques and neural network architectures, the overarching goal is to construct a robust model capable of accurately identifying handwritten words and characters across a diverse spectrum of handwriting styles and languages. By achieving this objective, the project endeavors to simplify and enhance the digitization process of handwritten documents. This advancement will not only improve archival practices but also enhance the searchability and accessibility of significant handwritten content across historical and contemporary domains. Through innovative approaches and rigorous methodology, the project seeks to contribute to the broader field of Handwritten Text Recognition, driving forward advancements in technology and paving the way for more efficient and accurate solutions in the future..

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

HAND WRITTEN TEXT RECOGNITION

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

This paper proposes a hybrid machine learning approach for URL phishing detection, combining supervised and unsupervised techniques. By leveraging features like domain information and employing models such as random forest and clustering algorithms, the system achieves high accuracy in identifying phishing URLs while minimizing false positives. This hybrid system offers robust protection against evolving phishing tactics, enhancing online security for users. Complementing the supervised approach, our system incorporates unsupervised learning techniques to uncover hidden structures within the data. Clustering algorithms, are utilized to group URLs based on similarity metrics derived from their feature representations. This unsupervised clustering aids in identifying anomalous patterns indicative of phishing behavior, thereby enhancing the system's ability to detect novel threats.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Hybrid machine learning, URL phishing detection, supervised learning, unsupervised learning, domain information, random forest, clustering algorithms, high accuracy, false positives, evolving tactics, online security

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Machine learning-powered facial age and gender estimation utilizes advanced algorithms such as Support Vector Machine (SVM) or K Nearest Neighbor (KNN) to estimate age of a person and gender from features of face in human face. It relies on extensive datasets of labeled facial images, which undergo preprocessing activites such as face detection, alignment, and feature extraction. These tasks ensure the extraction of relevant facial features like landmarks and textures. The SVM/KNN algorithms are used on curated dataset to learn decision boundaries separating different age groups and genders. This training process enables the models to make accurate predictions based on facial characteristics. The technology benefits from its ability to automate age and gender estimation tasks with high accuracy, facilitating applications in various domains such as security, marketing, and healthcare. However, challenges such as bias in training data and variations in facial expressions can affect the reliability of predictions. Additionally, privacy concerns related to facial recognition technologies are important considerations in its deployment.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Facial age and gender estimation, Machine learning, KNN Algorithm, Labeled dataset.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The "Animeflix" project outlines the design, development, and implementation of an Anime Streaming Website, leveraging modern web technologies such as TypeScript, HTML, CSS, and React for the frontend. The backend functionality is seamlessly integrated by directly importing data from an external API, ensuring a robust and dynamic streaming experience for anime enthusiasts.The frontend development is conducted using TypeScript, a statically-typed superset of JavaScript, to enhance code quality and maintainability. HTML and CSS are employed for creating a user-friendly and visually appealing interface, ensuring an immersive anime-watching experience for users. The backend architecture relies on importing data directly from an external API, eliminating the need for an independent backend server. This approach not only simplifies development but also ensures real-time updates and a vast library of anime content for users.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Animeflix, TypeScript, HTML, CSS, React, maintainability, user-friendly, visually appealing interface

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

SecureFaces stands out as a resolute guardian of online security in a time of digital interactions by fusing cutting-edge face authentication with a powerful Deepfake detecting tool. In this research, two powerful models--MesoNet for quick authentication and ResNetLSTM for increased accuracy--are shown. MesoNet and ResNetLSTM each address different user priorities. To ensure efficiency and data integrity, the system starts with an easy-to-use registration module that collects and safely stores facial data in MangoDB. Users can select between MesoNet and ResNetLSTM during the authentication process, depending on their unique needs, creating a customized identity verification strategy. SecureFaces is essentially an innovative cybersecurity solution that embodies trust, transparency, and adaptability in the digital sphere, going beyond simple facial authentication. This initiative acts as a beacon, guiding users through a trustworthy and safe authentication process as internet threats change

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Cybersecurity, Facial Authentication, Deepfake, Machine Learning, Deep Learning, Artificial Intelligence.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The internet's The "Face Mask Detection Using Convolutional Neural Network" project presents an innovative solution leveraging computer vision and deep learning techniques to automate the detection and categorization of individuals based on their adherence to face mask mandates. The system utilizes convolutional neural networks (CNNs) for robust feature extraction and classification, allowing real-time analysis of images or video frames to determine whether individuals are wearing masks and, if so, the specific type of mask. The project addresses the crucial need for efficient monitoring and enforcement of mask-wearing protocols in various settings, including public spaces, healthcare facilities, retail environments, and educational institutions. The application of this technology contributes to public health and safety by providing an automated, reliable, and scalable solution to ensure compliance with face mask guidelines, mitigating the spread of infectious diseases and enhancing overall safety measures in diverse sectors.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Face Mask Detection, Architecture, CNN

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Extractive text summarization using Latent Semantic Analysis (LSA) involves analyzing the underlying structure of a document by creating a matrix of term-document relationships. The TFIDF (Term Frequency-Inverse Document Frequency) vectorizer is employed to highlight important words in the document, assigning weights based on their frequency and uniqueness. Machine learning algorithms leverage these vectorized representations to identify and extract key sentences or phrases, forming the basis of the summary. Additionally, Deep Neural Networks (DNN) come into play, employing intricate layers of interconnected nodes to learn and understand complex patterns within the text. The DNN further refines the summarization process, enhancing the model's ability to capture nuanced relationships and context. This fusion of traditional ML and DNN approaches results in a powerful summarization system capable of distilling large volumes of information into concise, informative abstracts.Keywords--CNN, maximum pooling layers, dropout layers, softmax activation

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

LSA, DNN, TF-IDF Vectorizer

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The Sugar Cane is particularly crucial agricultural commodities in the world with a number of cuisines scattered across the globe, which are incomplete without it. In developing countries like India, Sugar Cane has spurred agriculture driven growth in the past century, when export of agricultural produce was the major source of foreign exchange. At times, the prices face a blow from the demand side, while at times facing drastic conditions on the supply side, owing to which, the prices of the commodity have seen a drastic fall. In such years, farmers often cannot afford the services of agricultural consultants for tasks such as of sicknesses of the leaves and addressing them at the earliest. The prescribed remedy is an inexpensive strategy which is easy use of image processing to detect leaf diseases in the leaves of Sugar Cane plants a way to streamline life for landowners in addition to consumers, since this would balance the prices at a median price. In this project, the affected leaves are captures as images using a camera. upon then, these photographs are adjusted. further using various methods and the key characteristics originate via them using various methods.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Sugar cane, Image Processing

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Approximately 39 percent of adult people on the planet are overweight. The fact presented above makes it clear how important and necessary exercise is. Exercise helps us maintain a healthy weight, a fit body, and a calm mind in addition to helping us lose weight. Regular exercise also keeps us active and improves blood circulation to keep our bodies in the same condition as when we used to visit the gym. It is expensive or not within everyone's reach to train under a trainer, visit the gym, take yoga courses, or both. Self-training is an additional option that provides pre-recorded yoga practice steps without any feedback. Without proper feedback about our postures, injuries can happen and it will do more harm than good and that's exactly where our project comes into play. These days, human position estimation is a widely pursued project in computer vision. The study of strategies and systems that retrieve an articulated body's stance is known as articulated pose estimation in computer vision. The course of determining the human body's location parts and joints in a given image is known as "articulated body pose estimation" in the context of this study. We study the many applications that we may put into practice with the data, which we received using a pre-trained posture estimation model called MediaPipe. Motion capture, gait analysis, anomaly detection, sign language recognition, and other uses are among them.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Human pose estimation, pose detection, pose estimation survey.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The sign language is used by people with hearing / speech disabilities to express their thoughts and feelings. But normally, people find it difficult to understand hand gestures of the specially challenged people as they do not know the meaning of sign language. Our project aims to develop a system forsign language recognition using MediaPipe ,LSTM, and Keras. The proposed system utilizes a webcam to capture real-time video input of a person performing sign language gestures. MediaPipe is used to extract and track the hand landmarks and their movements in the video stream. The features are then processed using LSTM, which is a sequence modeling technique that captures the hand gestures. Finally, a deep learning LSTM model implemented in Keras and trained to recognize the different sign language gestures. The system can potentially be used to assist people with hearing. Long Short-Term Memory (LSTM) neural network architecture to get this remarkable feat. When someone performs sign language gestures in front of a camera, the system instantly recognizes and interprets those gestures.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Animal species recognition, deep convolutional neural networks, transfer learning, camera-trap, KTH dataset.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Around 5000 million peoples have trouble hearing properly. While hearing aids can help somewhat, many struggle to understand speech when there's background noise. We've come up with a smart computer program that can filter out that noise while keeping speech clear. It works so well that it brings the clarity of speech is hearing aid users up to the level of people with thoda hearing. Here's how it works: We trained a computer program using a lot of recordings of speech with peecheka noise. Then, we made it even better by letting the computer figure out the best way to do this on its own. This program is really good at filtering out noise and letting you hear speech clearly, even if you're in a noisy place. It's much better than older methods that needed multiple microphones. And here's the exciting part: This program works fast, like in real time on a regular laptop. So, in a few years, we might be able to put it right into hearing aids, making zindagi good for millions of people with hearing problems.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Hearing, Noise, Train, Test.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

This paper makes use of deep learning techniques, specifically the ResNet-34 architecture, to present a novel method for estimating food volume and calories specifically for Indian cooked food items. The study looks at five common Indian dishes using computer vision to detect, categorize, and estimate volume of food. The procedure involves training a ResNet-34 model with a dataset that includes images of Indian foods such as biryani, curry, dal, roti, and samosas. Taking into consideration variations in preparation techniques and presentation styles, the model has been fine-tuned to accurately detect and classify these food items. Additionally, users can estimate the amount of food based on picture input thanks to the system's integration of volume estimation techniques. This feature is especially helpful for people who are watching their caloric intake or adhering to a diet. The outcomes of the experiments show how well the suggested method works for correctly recognizing Indian cooked labels, calculating their volumes, and estimating calories. The system's performance is evaluated across a broad range of datasets, showcasing its adaptability and reliability in diverse scenarios. All things considered, this work contributes to the field of food volume assessment and calorie estimation, especially as it relates to Indian cuisine, and offers a practical tool for monitoring nutrients and controlling diets.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Calorie estimation, Food classification, ResNet-34, Food Segmentation, Depth Network Training, Machine Learning, Health Monitoring OpenCV

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The " Voice recognition based I0T Home Automation System " is a cutting-edge loT project designed to revolutionize the way homeowners interact with their living spaces. This project, undertaken, has been crafted with the aim of bringing convenience and efficiency to everyday tasks through voice recognition technology and interconnected smart devices. By integrating voice commands through popular smart speakers, this system enables users to control a range of home devices, from lighting and climate control to security systems. With a strong focus on user privacy and data security, the project ensures that voice data is managed responsibly, and encryption measures are implemented for secure communication between the IOT hub and connected devices. The project boasts key features such as personalized voice commands, customization of routines, and real-time feedback. Users can create scenarios like "Movie night" that adjust multiple devices in a single command. A range of technologies, including cloud-based voice recognition services, IOT hubs, and end-device control mechanisms, are harnessed to create a seamless and intuitive experience for home owners. The "Smart Home Automation System" delivers numerous benefits, including enhanced convenience, energy efficiency, and improved quality of life. By simplifying daily tasks and promoting efficient energy use, it contributes to a more sustainable and comfortable living environment. As the project evolves, we anticipate implementing features to further enhance user experiences, such as integration with emerging smart devices and expansion of customization options.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

IOT, Smart Home, Voice Recognition, Automation

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

In our research, we harness the power of different computer algorithms to predict rainfall in various ecological zones of Ghana. We use data from the Ghana Meteorological Agency spanning four decades, from 1901 to 2015. We assess These algorithms' performance varies based on their accuracy in predicting rainfall, The speed at which they can operate these predictions, and their overall reliability. We find that Extreme Gradient Boosting, Random Forest, and Multilayer Perceptron algorithms excel across all three testing ratios, while K-Nearest Neighbour performs less effectively. Notably, Decision Tree proves to be the fastest in making predictions, but Multilayer Perceptron requires the most time. Our research provides valuable insights into the utilization of machine learning in tackling the complex rainfall prediction task in Ghana's diverse ecological regions.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Rain prediction using computers, Ghana weather data, Accuracy check, CNN, Reliability evaluation, Accurate results._

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

This project aims to create a sophisticated Rubik's Cube solver by blending human-designed algorithms with computer-based methods. By understanding the cube's structure and using efficient algorithms like Thistlethwaite, the solver analyzes the cube in real-time. It employs machine learning and deep learning techniques to enhance its efficiency. The software generates user-friendly 3D models and visualizations to help users understand the process better. Extensive testing ensures accuracy, with future plans to integrate physical devices for an even better user experience, potentially revolutionizing Rubik's Cube solving.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Rubik's Cube, Thistlethwaite, CFOP Algorithm, CUDA Architecture, OpenCV

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The internet connects the world, but also exposes it to numerous network threats. With the vast amount of information exchanged globally, ensuring the integrity and confidentiality of data has become increasingly challenging. Network security is essential in preventing easy breaches and unintended interference. One approach involves employing Intrusion Detection Systems, strategically positioned to monitor traffic from source to destination apps. However, balancing thorough screening with system efficiency is a concern. Integrating machine learning algorithms enhances flexibility and reliability in detecting and distinguishing between ordinary and malicious activities. The algorithms, logistic regression, Naive Bayes, K-Nearest Neighbour, and Decision Trees, are utilized in our research to optimize intrusion detection in Network Traffic Data, employing various evaluation methodologies to achieve the highest accuracy.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Cyber-attack, distance relay, graph theory, multi-agent system, distributed system, deep neural network.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Integrating The goal of this project, "Safety Helmet Wearing Detection Model Based on Improved YOLO-M," is to develop a model for determining whether or not people are wearing safety helmets. Improving safety monitoring in diverse settings is the aim. In order to develop a safety helmet detection system with an enhanced YOLO-M model, a computer must be trained to identify whether or not people are wearing safety helmets in images or videos. This entails utilizing data, modifying the software, teaching it to comprehend helmets, and verifying that it functions well. Once it functions properly, you can employ it in locations where you wish to verify that individuals are donning safety helmets.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

YOLO-M (You Only Look Once for Multi-Object Detection

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The goal of the project "Soil Analysis and Crop Recommendation Using Machine Learning" is to examine several approaches to soil analysis and crop recommendation. This model will forecast the greatest number of crops with a high degree of accuracy. It is essential to the survival and growth of the Indian economy.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Early Detection, Deep learning, CNN Algorithm, Image Processing.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

This paper presents a potential solution to break down communication barriers and promote inclusivity across various social and professional settings, bridging the gap between individuals with and without hearing impairments. The "Sign Language to Speech Conversion" and "Speech to Sign Language Conversion" initiative strives to create a real-time system that can translate Sign2Text & Text2Sign. The goal is to facilitate seamless two-way communication between individuals with hearing impairments and those without. The system will leverage advanced technologies CNN and RNNs to accurately recognize a wide range of gestures from sign language inputs. Furthermore, NLP models facilitate seamless translation of text. A key aspect is the real-time function of the system, minimizing delays in the translation process to enable instantaneous communication between ISL users and those relying on spoken language.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Spatial/temporal relationships, CNN, RNN, NLP

  License

Creative Commons Attribution 4.0 and The Open Definition