MUMBAI, India, Sept. 28 -- Intellectual Property India has published a patent application (202641113753 A) filed by Dr. Aryalekshmi B N; Ms. Noor Ayesha; Dr. L. C. Manikandan; Dr. Shobana D; Mr. Rajasekar A; and Mrs. Vishnu Gandhi V on September 23, 2026, for Intelligent Distributed Cloud Framework For Machine Learning-Based Image Analysis And Automated Feature Recognition.
Inventors include Dr. Aryalekshmi B N; Ms. Noor Ayesha; Dr. L. C. Manikandan; Dr. Shobana D; Mr. Rajasekar A; and Mrs. Vishnu Gandhi V.
The application for the patent was published on September 25, 2026, under issue no. 39/2026.
Abstract: This invention presents an innovative distributed computing platform for machine-learning-based image processing and automatic image feature identification. The system’s goal is to deal with situations where the image data is analyzed on various edge devices, local computers, accelerator platforms, and cloud systems. In traditional image-processing approaches, the ways of processing the image as well as the place of computing are fixed beforehand without observing the complexity of the incoming image or the real state of the computing resources of the system. In such cases, the fixed solutions may lead to excessive data transfers, delays in processing time, inefficient use of resources, and limited responsiveness in cases when the traffic or network conditions change. The inventive solution combines image acquisition and characterization, resource monitoring, the scheduling of tasks, machine-learning processing, feature extraction and recognition, and effectiveness estimation within a single processing system. Upon receipt of the image, the image analysis profile is made up based on some characteristics, such as image size, quality, texture, noise, density of properties, modality, and estimation of computation amount needed to analyze the image. By using this information, an advanced orchestration device comes to arrive at appropriate processing path. Hence, different processing nodes can handle tasks such as image preprocessing, feature extraction, machine-learning inference, recognition, and validation according to the needs of the image. Moreover, the system is flexible in that it allows for the use of different machine-learning models and configurations to meet different conditions of a particular image. In some cases, intermediate data such as regions of interest, feature representations, embeddings, segmentation output, etc., may be transferred instead of sending the entire image over the network again. A confidence-based validation mechanism further improves the processing operation. When the confidence associated with a recognized feature does not meet a selected condition, the system can initiate an alternative processing operation, such as applying another model, increasing image resolution, selecting another processing node, improving preprocessing, or allocating additional computational resources. Cloud-based processing may be used when connectivity is available, while edge and local processing can continue when cloud access is unavailable or degraded. Processing decisions, model information, recognized features, and confidence values may also be recorded to provide traceability and reproducibility.
Disclaimer: Curated by HT Syndication.
