ANPR system from Uncanny Vision

Traffic ANPR for highways: This uses Artificial Intelligence-based vehicle detection and classification technology, bundled with ANPR for traffic enforcement. The system automatically classifies the vehicles scanned and classifies the same based on the class of the vehicle providing very high accuracy. Traffic enforcement based on AI can be done using cameras alone or in combination with systems such as radars, traffic signal controllers, etc. Depending on the type of equipment used, traffic ANPR solution can be deployed in a manner suitable to the site. Traffic ANPR varieties are described below:

1. Deployment using Cameras alone for speed detection: In this case, a camera is a single sensor on the field that would be networked to a server with the AI engine to perform traffic ANPR. Use cases include spot speed detection, average speed detection from two cameras, vehicle duration detection between two cameras, and parking violation detection. Ideal for speed upto 40 kmph.
2. Red light violation detection: The camera can be integrated with traffic signal controller at junctions. Based on the signal sent by the controller, the camera can be triggered to capture video and the stream can be sent for detection of violations such as red light jumping, vehicle occupying area beyond traffic lights etc. Non-invasive method detects red light violation. The ANPR of the violating vehicle can be sent to endpoints depending on the integrations.

Traffic ANPR in conjunction with Radar:

Solely camera-based speed detection systems perform better depending on the clarity of the image captured at different vehicle speeds. Using a Radar+ANPR system allows an increase in the range of speed violations captured i.e detection of speed at longer distances. The radar component would capture the speed of multiple vehicles at different ranges and based on the set trigger send an image for ANPR detection. The metadata from the radar would be integrated with the camera. Radar offers a wider range of detection in terms of distance and multi-vehicle tracking and higher accuracy for speed detection. The camera provides visual evidence for collaboration and for AI-based ANPR processing. Violations which can be detected are over speeding, wrong direction, helmetless driving by 2-wheeler and triple riding in 2-wheelers.

Using a radar + ANPR solution, detection of helmetless riding can be provided to authorities for enforcement purposes. The solution can be mounted on gantries or on interceptor vehicles.

Stop and Go ANPR Solutions at tollbooths or checkpoints:

This performs at locations where the flow of vehicles occurs in a stop and go manner. In this use case, a camera is placed at checkpoints such as toll gates, weighbridges and AVCC checkpoints to detect the vehicles and perform number plate recognition. Based on the license plate, the following objectives can be achieved:

1. Digital record of vehicle transactions occurring at toll plazas based on number plate can be stored and used for investigative or verification purposes
2. Combining ANPR with other devices such as FasTag to get complete holistic information about the transaction vehicle
3. Digital records of all vehicles passing through the gate/checkpoint can be stored as an alert
4. Alert for blacklisted vehicles passing by toll booth

Pedestrian detection:

AI based analytics to detect person seen in the region of Interest. Can be used to spot Pedestrians in given region of interest in camera view. This analytics would help in spotting people at different sections of the highway, road crossing etc. The analytics can be use monitor scenarios of jaywalking, person intrusion in hot spots etc and take appropriate action.

Stationary Vehicle detection (due illegal parking, accident etc)

Detection of stationary vehicle in a defined camera region of interest. The AI can send an alert if a vehicle is seen in the region of interest beyond a defined period of time. The time can be configured to distinguish between an illegally parked vehicle and vehicles stopped due to accidents. ANPR would be done on best effort basis if the number plate is humanly readable in the frame used for inferencing.

Unusual traffic snarls (High number of vehicles in region of interest):

To monitor and reduce road congestion we can use AI based solutions to detect the count of vehicles in a region of interest. If the number of vehicles is more than a given threshold an alert can be sent to respective authorities for further action. Image of vehicle detection and counting shown below.

Detection of Boulder and cattle on roadways

Certain stretches of the roadways would be prone to blockages due to the presence of animals or boulders due to landslides etc. Our smart AI solutions allows the detection of cattle and large rocks (boulders) present on the road in a given region of interest. The alerts sent to an endpoint can be used by authorities to pursue defined SOP’s to enable unhindered pathways.