Intellectual Property India Publishes Patent Application for Integrated Vision-Based Vehicle Detection And Timeseries Forecasting System For Parking Occupancy Prediction

MUMBAI, India, July 13 -- Intellectual Property India has published a patent application (202631077587 A) filed by C. V. Raman Global University on June 23, 2026, for Integrated Vision-Based Vehicle Detection And Timeseries Forecasting System For Parking Occupancy Prediction.

Inventors include Soumyaranjan Sahoo; Saquib Khan; Pratik Verma; Amlan Prashad Sahoo; and Dr. Ram Chandra Barik.

The application for the patent was published on July 10, 2026, under issue no. 28/2026.

Abstract: The present invention discloses an Integrated Vision-Based Vehicle Detection and Time-Series Forecasting System for Parking Occupancy Prediction, which provides an end-to-end automated pipeline for real-time vehicle detection and short-horizon occupancy forecasting without requiring per-slot physical sensor hardware. The system comprises a data acquisition layer employing CCTV or IP cameras, a preprocessing module applying Contrast Limited Adaptive Histogram Equalization (CLAHE) to enhance video frame quality under adverse illumination conditions, a fine-tuned YOLOv8m object detection engine achieving approximately 97.9% precision and 96.9% recall, an occupancy computation module employing BoT-SORT and ByteTrack multi-object tracking algorithms, a Bidirectional Long Short-Term Memory (Bi-LSTM) forecasting module generating probabilistic occupancy predictions up to thirty minutes ahead with an R-squared value of approximately 0.988, a centralized data storage module, and an API communication layer delivering real-time and forecasted occupancy data to operator dashboards, mobile applications, digital signage, and smart-city integration platforms. The invention uniquely couples live visual detection outputs directly with the forecasting module in a single automated pipeline, overcoming the fragmentation of prior art systems that address detection or forecasting in isolation. The sensorless architecture leverages existing CCTV infrastructure, eliminating installation and maintenance costs associated with per-slot hardware while enabling scalable deployment across commercial, municipal, institutional, and transportation hub parking environments. An online adaptive learning mechanism ensures sustained forecasting accuracy under evolving demand patterns, and a federated deployment option supports privacy-preserving cross-facility model improvement.

Disclaimer: Curated by HT Syndication.