MUMBAI, India, Aug. 24 -- Intellectual Property India has published a patent application (202641098649 A) filed by Avs Engineering College on August 14, 2026, for System And Method For Deep Learning-Based Image Recognition Using Convolutional Neural Networks.

Inventors include Mrs. A. Anandhi; Dr. K. Ambika; Mr. P. Vallarasu; Ms. Priyadharshini Settu; Mrs. P. Deepalakshmi; Mrs. K. Vidhya; Mr. P. Thamizharasu; Ms. S. Ramya; and Ms. S. Bhoomika.

The application for the patent was published on August 21, 2026, under issue no. 34/2026.

Abstract: ABSTRACT SYSTEM AND METHOD FOR DEEP LEARNING-BASED IMAGE RECOGNITION USING CONVOLUTIONAL NEURAL NETWORKS The present invention discloses a system and method for deep learning-based image recognition utilizing convolutional neural networks and reinforcement-driven optimization. The system comprises a data preprocessing unit configured to normalize, augment, and filter raw image inputs, followed by a feature extraction module employing convolutional layers, residual connections, and attention mechanisms to generate hierarchical feature representations. A training and optimization engine integrates supervised learning with transfer learning from pre-trained models, enhanced by reinforcement learning strategies that dynamically adjust hyperparameters to improve recognition accuracy and reduce overfitting. An inference layer provides real-time classification, object detection, and anomaly recognition outputs, enabling deployment in critical domains such as healthcare diagnostics, autonomous navigation, and cybersecurity threat detection. The deployment and scalability unit supports cloud, edge, and embedded platforms, incorporating model compression techniques including pruning, quantization, and knowledge distillation to reduce computational overhead. Novelty lies in the integration of reinforcement-driven optimization with CNN architectures, enabling adaptive learning and interpretable outputs. The invention delivers improved diagnostic precision, enhanced safety, and robust threat detection, establishing a scalable and efficient framework for next-generation image recognition systems.

Disclaimer: Curated by HT Syndication.