Image and Video Analytics

Mayachitra Inc. offers a suite of solutions for object detection, recognition, and tracking, activity detection and analysis, and user interactions targeted for data smog reduction in aerial video sensor feeds.

Mayachitra is presenting its video analytics technology at Navy Forum, April 3-5 2017, National Harbor MD, (N14A-T008 Mayachitra.pdf)

Object Detection and Recognition in overhead video streams

Exorbitant volumes of collected video and imagery data require advanced technology to analyze the content in real time. Low resolution quality of operational data, size of objects of interest, view occlusions, and crowded scenes degrade the performance of state-of-art region saliency and object recognition approaches when applied to overhead sensor data.

Mayachitra Inc. provides state-of-the-art computer vision platform dedicated to meeting the needs of defense analysts for over-head surveillance videos from multiple cameras. The system brings together the latest in deep neural networks and computer vision to enable the defense analyst to monitor and analyze terabytes of streaming overhead videos.

We have developed a technology to automatically detect and recognize multitude of objects of potential interest providing an object recognition decision with a high level of confidence. Our technology employs state-of-art deep learning and bio-inspired methods to efficiently and effectively detect small objects in overhead noisy (crowded occluded) videos, where state-of-art models fail.

Objected detection in overhead video examples

User Interaction and Activity Modeling

The technology comes with the suite of the support modules that enhance user interaction and modeling. The focus is to allow analysts to let the system take over the tedious and error prone task of tracking objects of interest across multiple videos that are distributed across large swaths of space and time. The system alerts the analyst on specific events relating to those objects, and learns from its mistakes by seeking active feedback from the analyst and will use the latest techniques in machine learning to minimize user feedback while maintaining precision and accuracy.

System automatically labels and tracks object of interest

The system alerts the analyst when the confidence of tracking falls, and asks for an active feedback

Mayachitra Inc. team has done extensive work on User Interaction and Video Activity, see demo for more details.

This work has been supported by NAVAIR and the Office of Naval Research.