Dashcam Intelligence
& Connected Analytics
Overview
The Astroc Nebulae Automotive Division brings real-time intelligence to connected vehicle camera feeds through cloud-based processing. From a single dashcam stream, the platform simultaneously identifies vehicles, reads number plates, estimates inter-vehicle distances, counts pedestrians, recognises traffic signs, and extracts text from roadside boards. All of this runs on the Astroc Vision System, our proprietary detection and analytics engine built for edge and cloud deployment.
What the System Detects
The Astroc Vision System processes full HD dashcam footage at 29 frames per second and can identify over 23 categories of objects. These include cars, buses, trucks, motorcycles, bicycles, pedestrians, traffic lights, animals, and various roadside objects. Every detection is tagged with a confidence score and an estimated distance from the camera, calculated using a ground-plane perspective model that has been specifically calibrated for wide-angle dashcam lenses.
Number Plate Recognition
An integrated OCR engine reads vehicle registration plates directly from the dashcam feed. Each plate read is linked to the vehicle it belongs to, so the output shows not just the plate number but also whether it was on a bus, car, truck, or motorcycle. The system is tuned for Indian registration formats and handles partial reads, oblique angles, and varying lighting conditions. Every read comes with a confidence score so operators can gauge reliability at a glance.
Sign and Board Reading
The platform goes beyond standard traffic sign detection. It uses text-region detection and contour analysis to find and read shop names, advertising boards, road signs, and information displays visible in the camera feed. This makes it possible to automatically catalogue roadside infrastructure and commercial signage across an entire driving corridor without any manual effort.
Distance Estimation
Rather than relying on object size alone, the system uses a ground-plane perspective model to estimate how far each detected object is from the vehicle. Objects near the bottom of the frame are treated as close (typically 1 to 3 feet), while those higher up are progressively further away. This approach is far more reliable with wide-angle dashcam lenses than traditional focal-length methods, and the output is shown in feet for immediate, practical use.
The Nebulae Dashboard
All analytics feed into the Astroc Nebulae Dashboard, an intelligence platform designed for operational clarity. It presents live KPI tiles showing total detections, vehicle and pedestrian counts, plate reads, and average spacing. Below the metrics, the annotated video plays with bounding boxes, distance labels, and plate highlights overlaid in real time. Temporal charts, confidence radars, detection heatmaps, and scrollable data tables round out the interface.