Journal of Soft Computing and Computational Intelligence (p-ISSN: 3107-4855, e-ISSN: 3048-6610)

Implementation of an AI-Based Architectural Image Captioning System for Visual Accessibility

Hima R — 2026-05-30

The widespread availability of architectural imagery across digital platforms, combined with the lack of automated description tools, creates a significant accessibility barrier for visually impaired users and multilingual communities. Architectural images often contain complex structural details that are difficult to interpret without domain-specific knowledge, making manual captioning impractical at scale. This paper presents an AI-assisted architectural image captioning system designed to bridge the gap between architectural visual content and accessible language. The proposed system integrates an InceptionV3-based Convolutional Neural Network (CNN) for robust visual feature extraction and a custom Long Short-Term Memory (LSTM) decoder, trained on domain-specific architectural datasets, to generate accurate, context-aware captions. To enhance accessibility, a multilingual Text-to-Speech (TTS) module is incorporated, supporting English, Hindi, and Kannada, enabling users to receive audio descriptions in their preferred language. Additionally, a user-friendly graphical interface is developed using Tkinter, allowing real-time image upload, caption generation, and audio playback. The implementation demonstrates that combining domain-specific deep learning techniques with multilingual audio output can effectively transform architectural images into meaningful spoken descriptions. The system is lightweight, scalable, and operates efficiently on standard hardware without requiring cloud-based infrastructure. This work highlights the potential of artificial intelligence in improving accessibility, supporting architectural education, and enabling inclusive interaction with visual content for diverse user groups.

A Review of Brain Tumor Detection in MRI Using Hybrid Fuzzy C-Means Clustering and Support Vector Machine Techniques

Kamini Panthi — 2026-05-30

The heterogeneous and complex nature of the brain tumors makes its detection and classification important to the accurate diagnosis, prognosis and treatment planning. The tumors are classified on the basis of their origin as either primary or secondary and on the basis of behavior as either being benign or malignant and each of these classifications has an effect on the clinical presentation and treatment options. The location of the tumor in the brain also has an additional impact on the symptoms, accessibility of surgery, and treatment. The modality of choice is Magnetic Resonance Imaging (MRI), which has all the advantages of high soft tissue contrast, multiplanar imaging, and functional applications, including diffusion-weighted imaging, perfusion MRI, and magnetic resonance spectroscopy. This work is based on the suggestion of a hybrid system, with Fuzzy C-Means (FCM) clustering used for segmentation and a Support Vector Machine (SVM) used as a classification tool. FCM is used to assign probabilistic pixel values to represent overlapping tissue properties, and SVM is used to classify segmented regions using features of intensity, texture, shape, and edges. The hybrid method improves the ability to detect and distinguish the types of tumors and helps radiologists identify and plan treatment much earlier and monitor brain tumors through ongoing procedures.

An Integrated Artificial Intelligence and Machine Learning Approach for Real-Time Disease Prediction

Subhasini Shukla — 2026-05-30

The rapid advancement of artificial intelligence has played a massive role in the medical and healthcare fields. Globally, disease prediction has been indispensable. Artificial Intelligence (AI), Machine Learning (ML) and Deep Learning (DL) technologies have had a significant impact on predicting or detecting the symptoms that can cause a huge loss in the future. Nevertheless, conventional approaches to disease comparison are resource-intensive in terms of both time and cost, whereas machine learning integrates deep learning techniques for more efficient analysis. Deep learning is a branch of machine learning that employs artificial neural networks to mimic how the human brain processes information, enabling it to identify patterns from large volumes of data. There is an intelligence system that analyses data to come up with valuable information for prediction purposes. This application focuses on the future possibilities and challenges, and harnessing technologies to develop pioneering public health solutions. The computers have the ability to perform calculations without any pre-programming through the help of machine learning technologies. Machine Learning uses the ideas of synthesis and induction to improve computers. It is employed in a variety of fields, especially bioinformatics and the diagnosis of diseases. Chronic illnesses, particularly cardiovascular diseases (CVDs), are among the leading causes of death worldwide, making it essential to develop accurate and timely prediction systems for early diagnosis and effective preventive care.