Volume 14 | Issue 4 |

 Abstract

Fungi act as decomposers by breaking down organic matter in nature, and many form beneficial relationships with plants in the soil. However, some fungi develop parasitic relationships with plants, such as rusts and leaf spots. These disease-causing fungi obtain their nutrients from the plants they infect. They cause significant damage, leading to symptoms like wilting, scabs, moldy coatings, rust-like patches, and rotting tissue. In more developed areas, plant diseases are estimated to reduce crop yields by about 10% annually, while in less developed regions, the yield loss due to diseases can be more than 20%. Forests provide essential resources such as timber, fuel, and fodder, and also help conserve soil and water, regulate climate, support wildlife, and offer recreational and aesthetic benefits to humans. There is a strong connection between plants and their environment. In forest nurseries, various diseases like damping-off, root rot, stem rot, leaf curl, wilt, canker, rust, and decay can be observed. In this study, we have examined specific and widespread plant diseases in forest nurseries of Sonebhadra, Chandauli, and Mirzapur districts in Uttar Pradesh, India, caused by fungal pathogens. Many soil-dwelling fungi can live as saprophytes, completing part of their life cycle in the soil, and are known as facultative saprophytes. Pathogenic conditions can result from both biotic and abiotic factors. Fungi reproduce both sexually and asexually through the production of spores and other structures. Spores can travel long distances through air or water, or they may be present in the soil. Fungi have been classified into four groups by Alexopolous and Mims (1979) as Zygomycetes, Ascomycetes, Basidiomycetes, and Deuteromycetes. In my research, most plant-disease-causing fungi belong to the Ascomycetes and Basidiomycetes groups. These have been studied because there has been very little previous research conducted in different forest areas.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

FUNGI, DECOMPOSER, FOREST, DISEASE, SPORES

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

"One of the first conditions of happiness is that the link between man and nature shall not be broken." - Leo Tolstoy. Keywords: Environmental Education (EE) Package, Environment Awareness, Environment Behaviour, Value Action Gap, Quasi-Experiment, Pro-environment action This research investigates the effectiveness of a specialised Environmental Education (EE) Package in enhancing environmental awareness and behaviour among secondary school students in Anand, Gujarat, with a specific focus on the "value-action gap," where knowledge often fails to translate into sustainable habits. Utilising a quasi-experimental pre-test/post-test design with 100 students, the study compared a traditional lecture-based control group against an experimental group exposed to a multi-sensory 15-day intervention. This EE Package integrated multimedia presentations, hands-on activities, and cultural storytelling from the Ramayana and Bhagavad Gita to foster stewardship. The results demonstrated a statistically significant improvement in both awareness and pro-environmental actions in the experimental group, which significantly outperformed the control group. Furthermore, a strong positive correlation of 0.91 was found between awareness and behaviour, indicating that the package effectively translated ecological knowledge into practical action regardless of student gender.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Environmental Education (EE) Package, Environment Awareness, Environment Behaviour, Value Action Gap, Quasi-Experiment, Pro-environment action

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

There is a growing need for dependable and effective charging infrastructure as a result of the quick uptake of electric cars (EVs). However, finding available charging stations, scheduling charging times, and obtaining flexible battery exchange choices are frequently difficult tasks for EV customers, which causes delays and annoyance. The suggested system makes use of real-time data integration and sophisticated location-based services to give consumers precise information about the availability of charging stations and battery swap stations. It provides EV users with an easy-to-use mobile application that lets them look for charging stations and battery exchange locations in their area, browse station information, check slot availability, and make reservations ahead of time. Additionally, the system has a secure payment gateway that enables customers to reserve charging or battery exchange times and complete transactions with ease. Wait times are reduced and station utilization is maximized using an effective slot management algorithm. The suggested system greatly improves user convenience and helps create a more sustainable and orderly EV charging infrastructure by incorporating safe payment processing, battery exchange possibilities, and real-time notifications. The system architecture, implementation difficulties, and possible advantages for EV consumers and charging station operators are covered in this article.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Electric Vehicle (EV), Charging Station Locator, Mobile Application, Slot Booking System, Smart Charging Infrastructure, Location-based Services.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Large language models (LLMs) have emerged as powerful tools for software engineering tasks including code summarization, bug detection, and architectural analysis. However, integrating these models with retrieval-augmented generation (RAG) to ground responses in trusted evidence presents significant challenges for enterprises and startups. Most existing LLM-based systems depend heavily on cloud infrastructure, introducing critical concerns related to data privacy, intellectual property protection, regulatory compliance, and operational costs. This is particularly problematic for organizations managing proprietary source code where data sovereignty is paramount. This literature survey comprehensively reviews recent research on offline retrieval-augmented AI systems for secure enterprise code intelligence, analyzing LLMs, RAG frameworks, hybrid retrieval mechanisms, secure knowledge storage, and lightweight deployment architectures. The survey encompasses works from 2023-2025 across IEEE, Elsevier, and Springer, examining deployment feasibility, architectural paradigms, and enterprise applicability. Our analysis reveals critical research gaps: most approaches remain cloud-dependent, lack offline deployment capabilities, treat source code as unstructured text without understanding repository structure and dependencies, and demonstrate insufficient integration of hybrid retrieval in private environments. Furthermore, unified offline knowledge base architectures that seamlessly integrate semantic embeddings, keyword indexing, and structural metadata remain largely unexplored. These observations motivate the need for a secure, offline, cost-effective, and startup-friendly code analysis system. We identify promising research directions toward lightweight generative models, secure retrieval ecosystems, parameter-efficient fine-tuning, and hardware-aware inference optimization that enable next-generation AI-powered code intelligence platforms capable of operating entirely within private organizational infrastructure while maintaining high reasoning accuracy and developer productivity.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Large Language Models, Retrieval-Augmented Generation, Offline AI Systems, Secure Code Analysis, Private Knowledge Base, Lightweight Deployment, Enterprise Software Engineering, Vector Databases, Hybrid Retrieval

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Sign language serves as a vital mode of communication for individuals with hearing and speech impairments, yet its widespread adoption is limited by the lack of mutual understanding between signers and non signers. A sign language translator aims to bridge this gap by converting hand gestures into meaningful text or speech in real time. The system typically integrates computer vision techniques for gesture detection, machine learning models for classification, and natural language processing for refining output into grammatically coherent sentences. By combining image acquisition, feature extraction, and predictive modelling the translator provides an accessible communication channel that promotes inclusivity and reduces barriers in education, healthcare, and everyday interactions. This work highlights the potential of gesture to text systems to enhance accessibility, while emphasizing modularity, scalability, and adaptability for diverse sign languages and user contexts.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Sign Language Recognition , Computer Vision , Random Forest Classifier , Natural Language Processing , Gesture-to-Text Translation, Accessibility Technology , Real-Time Communication

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Rapid urbanisation and increasing urban population density result in cities' expansion, often intensifying economic and social inequalities. This emphasis needs the adoption of long-term sustainability goals. Current literature highlights the urgent need for policy interventions at the implementation level to achieve integration of Sustainable Development Goals (SDGs) and long-term sustainability. To facilitate the assessment of sustainable development, green rating systems have been introduced globally. The study uniquely examines the LEED for cities and communities (LCC) framework from social, environment and economic perspectives. The LCC framework encompasses nine different categories. Each category revolves around specific themes that can be addressed by the application of specific strategies to earn a credit point. Each category is evaluated against its performance score, based on the strategies adopted and its mapping to the Triple Bottom Line (TBL) sustainability framework. The key pillars of TBL structure are Social equity (People), Economic viability (Profit) and Environmental responsibility (Planet). The study aims to understand how each strategy across nine categories contributes to achieving SDGs, aligning with the TBL structure. This study adopts a qualitative research technique using a systematic coding process to identify themes related to LCC strategies. The assessment indicates that LCC is a comprehensive framework that fosters urban development, aligning with global sustainability goals. Sustainable development goals like health and wellbeing, industry innovation and infrastructure, Sustainable cities and communities and climate action show the highest linkages to the LCC framework. A key recommendation of this study is to strengthen efforts toward enhancing awareness and understanding of economic, social, and environmental values among stakeholders, thereby promoting healthier and more environmentally responsible lifestyles. Policymakers and planners, in particular, are encouraged to incorporate the strategies and indicators proposed within the LCC framework into decision-making processes.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Sustainable urban development, Framework analysis, SDGs

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Emergency response in India is often slowed down by a reliance on manual work and a scattered network of different helplines, like 100 for Police, 101 for Fire, and 108 for Ambulance. This fragmentation leads to lost time and inefficient use of resources when people need them most. This paper introduces ResQ AI, a smart, web-based platform designed to streamline emergency response. By using machine learning, the system automatically figures out how severe an incident is and flags potential false alarms. The platform brings together three key groups--Citizens, Volunteers, and Administrators--giving each a tailored dashboard built with a modern React and Python-based setup, supported by a MongoDB database. We used a Random Forest Classifier trained on 1000 records to reach 97% accuracy in categorizing emergencies from "Low" to "Critical." Meanwhile, an Isolation Forest model works in the background to catch suspicious or duplicate reports as they happen. The platform is equipped with helpful features like automatic GPS location, interactive maps, and secure login protection. Our testing shows that ResQ AI significantly cuts down the time spent on coordination and makes responding to emergencies much more efficient than the traditional systems we use today.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Emergency Response, Machine Learning, Random Forest, Isolation Forest, Flask, React, MongoDB, Severity Prediction, False Alarm Detection, Real-time Coordination

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

This study examines the measurable impact of Artificial Intelligence (AI) tools on the Academic Performance of undergraduate students in the Imphal Valley of Manipur. Unlike earlier research that primarily focused on perceptions, this investigation emphasizes outcomes such as grades, study habits, and assignment quality. Using a Descriptive survey method, data were collected from 40 students at Imphal College and G.P. Women's College using structured questionnaires that combined quantitative and qualitative responses. Results show that AI tools - particularly ChatGPT, Google Gemini, and Grammarly - are widely used, improving exam preparation, time management, and the quality of academic work. At the same time, challenges such as overdependence, reduced originality, and weakened critical thinking were identified. The study concludes that AI can enhance academic performance when used responsibly, serving as a supportive companion rather than a replacement for creativity and independent effort. It recommends that educators and institutions promote AI literacy and ethical integration to ensure balanced learning, where technology complements human abilities and fosters confident, thoughtful, and independent learners.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Keywords: Artificial Intelligence (AI), Educational Technology, Academic Performance, Critical Thinking, Responsible Integration

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Brain tumors represents a significant medical concern andtheirearly and accurate detection plays very importantrolein improving treatment outcomes. Among the available diagnostic techniques,Magnetic resonance Imaging (MRI) is the most common method used to detect braintumors , but analyzing MRIs cans manually takes a lot of time and depends heavily on expert doctors. This manual process may raise the errors and delay in the process. This study proposes an automated framework for brain tumor detection using Deep learning models which is useful in detection of tumor cells and uneven cells using the proposed system combines two models U-Net and Efficient NetB5 . Where U-Net is used for segmentation that includes highlights the exact tumor location in MRI image by separating it from Healthy brain tissue. After segmentation the identified cancer region under goes to Efficient netB5,which is a deep learning model designed for classification. Efficient NetB5 classifies the tumor belongs to which type such as Glioma, Meningioma or Pituitary tumor. The proposed combination model is not only able to detect where tumor is located but also identify what type of tumor it is . Tests on standard MRI datasets show that this approach improves accuracy , reduce errors and speed up the treatment process. Our project is a additional scope for professionals to analysis the patient condition and provides a reliable computer based tool for faster and better treatment planning

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Efficient-Net; U-net; Accuracy; Glioma; Meningioma; Pituitary; Segmentation; Classification Union; Neuro-Diagnostic; Magnetic resonance Imaging(MRI)

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Every year, road traffic accidents claim hundreds of thousands of lives worldwide. A significant portion of these deaths occurs not from the severity of the crash itself but from the delay in emergency response. This paper presents an end-to-end solution to this problem: an automated accident detection and response system built using the YOLOv8 deep learning model and SORT algorithm. Our platform continuously monitors live video streams from multiple sources, including local CCTV cameras and RTSP network feeds. The system detects accidents in real-time and logs the incident to a MongoDB database. Immediately after an accident is confirmed across at least three consecutive frames, the system dispatches an alert with the location coordinates, an image of the accident, and the severity to emergency responders. A React.js and Next.js web dashboard gives supervisors a clear view of the live streams. In tests on highway collision videos, the system achieved an accuracy of 87 to 92 percent and reduced the incident dispatch time from several minutes to under five seconds.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Accident Detection, YOLOv8, SORT, Computer Vision, Emergency Response, OpenCV, Flask, React.js, Intelligent Transportation Systems, Deep Learning, Object Tracking, Real-Time Video Analytics.

  License

Creative Commons Attribution 4.0 and The Open Definition