Challenge: Reliable and safe pedestrian crossing.

Solution: Client was using motion sensors & low frequency radar for detecting people on the side of the road for pedestrian crossing. However, this is not completely reliable as sensors might detect any object as human and change the traffic light. We designed a vision based system, it detects the person and allows them to cross. It also records video footage to support evidence w.r.t. accidents and over speeding. 180 video streams (1080p @30 FPS) are processed in realtime by a single T4 server (4*Tesla T4).

Benifits: Reduced cost, increased accuracy, in turn road safety.

Challenge: Improve traffic management through early detection of oncoming vehicles at intersections.

Solution: SmartCow developed a vehicle collision avoidance system. SmartCow used Sentinel that runs multiple neural networks in parallel for applications like image classification and object detection. All in an easy-to-use platform that runs in as little as 20 watts to detect oncoming vehicles. The edge system is connected to high beam traffic signal system (regulate the passage of vehicles and pedestrians). SmartCow installed at few corners as pilot project. These traffic lights are much more effective than mirrors used at the turning of the road, so the maintenance is easy and affordable.

Benifits: Accidents are reduced by 60% in the areas where the SmartCow's Sentinel is installed. Looking at the results, government is eager to install more systems.

Challenge: Cost optimizing and improving bandwith

Solution: This project requires Automated Number Plate Reading (ANPR /ALPR) on thousands of RTSP cameras a combination of H265 & H264. SmartCow's Cerberus and SuperCam is used for this project; a single unit can manage 30 cameras with a storage capacity of 2TB data for 2.5 years

Benifits: Better traffic and public management, reduced crime in real time, easy scale up