AN INTELLIGENT DEEP LEARNING FRAMEWORK USING YOLO-BASED CNN–LSTM–ATTENTION NETWORKS FOR BIOMEDICAL IMAGE ANALYSIS AND MOLECULAR PATTERN RECOGNITION

Abstract

Background: The significance of intelligent monitoring systems in railway safety has grown, particularly in detecting animal intrusions and dangerous objects on tracks. Advances in deep learning, computer vision, and biomedical image intelligence have facilitated automated biological pattern recognition and large-scale image analysis, paving the way for developing effective surveillance systems that enhance public safety and environmental monitoring. Objectives: This research introduces an intelligent deep learning framework that combines YOLO-based object localization with a CNN–LSTM–Attention architecture for automated railway surveillance, detection of animal intrusions, recognition of hazardous objects, and analysis of biological patterns. The framework utilizes advanced image preprocessing methods, such as image enhancement, noise reduction, normalization, and region-of-interest (ROI) extraction, to enhance visual information quality across various environmental conditions. Methods: The YOLO detection module precisely identifies animals, hazardous objects, and railway track areas from surveillance video sequences. The convolutional neural network extracts distinct spatial features, while the long short-term memory network captures temporal relationships from successive video frames. Additionally, the attention mechanism selectively highlights biologically significant regions and hazardous objects, enhancing classification accuracy and decision-making in intelligent railway monitoring. Results: The proposed framework uses computational feature optimization and statistical learning techniques to effectively represent high-dimensional visual features. Experimental results show superior performance in accuracy, precision, recall, F1-score, and robustness compared to traditional deep learning methods. The system efficiently detects track intrusions, hazardous objects, and animal movements, reducing false alarms and improving real-time surveillance capabilities. Conclusion: The artificial intelligence-driven framework offers an efficient solution for intelligent railway surveillance, detection of hazardous objects, monitoring of animal intrusions, and biological image intelligence. It supports future applications in intelligent transportation systems, bio-inspired computer vision, automated safety monitoring, environmental surveillance, and next-generation smart railway infrastructure.