Computer Vision solutions across domains

Computer Vision (CV) is all about the machines to see like humans do and process and interpret what was seen, like humans do. The processing can happen at the machine side (if it has an in-built processor) or can happen in the connected device (servers in cloud or in-premise). The seeing part off course is via a camera (CCTV, Mobile-Phone, Dashboard, Robot eye, or a Satellite).

Coforge has been using Computer Vision capability for providing solutions across domains. Below table provides a high level view of some of the task that are auto mated using computer vision technology.

Slno CV Task Description
1 Object Detection Combination of object classification (a car) or object recognition (a Fiat) with object location (where in the image is the object)
2 Caption Generation Understanding and recognising objects in scene. Ex: Man kicking football
3 Recognising Human Pose Predicting person movements. Ex: Human fall or violence detection
4 Object tracking An entity is followed through the scene
5 Gesture Recognition To recognize the gestures made by human hand

Case Example 1: For a large transportation company in SE Asia, we provided a computer vision solution to automatically count and measure different sizes of wood. An instance segmentation solution was built using Mask R-CNN deep learning model. This resulted in identifying and tracking any fraudulent activities happening in logistics.

Case Example 2: For a large Asset Management Firm we developed an AI solution to detect oil storage tanks and also estimate the volume of oil stored in the tank, from geo spatial images. A Faster R-CNN based model used to build the solution. This helped in understanding the impact on industries dependent on Oil/Petrochemicals

Case Example 3: In one of the customer facing contactless kiosks of a leading airport we developed Human hand gesture recognition system. Solution was to count the number of fingers held up to seek answers like total number of check-in baggage. The contactless solution were part of safety measures for people in times of a global pandemic (COVID-19).