Vision AI.
So humans can be humane.
We design and develop deep large-scale AI and ML vision solutions for Enterprises, and Defence & Security Organizations.
COGNITION
COMPREHENSION
COMPRESSION
INTEGRATED AI & VISION STAGES
An integrated AI approach where each prediction stage links up to a richer prediction stage, in a single cycle.3D Reconstruction & HPE
The predicted 3D world provides egocentric semantics for scene & Pose comprehension.
MOTION ESTIMATION & FEATURE EXTRACTION
Our deep neural layers for feature extraction performs local & global motion estimation.
Video & Optical Flow
Optical Flow is an encoding byproduct; we steal it at video encoding stage.
ACTION/SCENE RECOGNITION
LSTM Models then predict actions and motions, for "in-frames" too.
EMOTION & ZERO SHOT DETECTION
Faces go to Emotion detection models, newly seen features to ZERO Shot detector.
COMPRESSION & GLOBAL MODEL
All stages till now help create heavily compressed video and a fully segmented Global model.
OUR SOLUTIONS
Virtual Drone
A 3D VR environment with 100s of integrated cameras for large facilities like an airport, tech parks, etc.
Video Compression
Up to 96% compression on H.264 using AI-based motion and scene understanding.
Large Scale Face Detection
People count and face detection with more than 95% accuracy for large crowd and rallies.
Multi-Camera Tracking
Detect and Track same objects in multiple cameras real-time or offline videos.
Facial Super Resolution
Specially trained for Indian faces, achieve up-to 5X facial superresolution.
Colorization
BnW or Near IR imaging to RGB conversion. Camera intrinsic modeling for targeted conversion.
HPE & Facial Pose
Multi-Human Pose combined with 3D Facial Pose estimation for quantifying engagements.
Action Recognition
LSTMs and RNNs for training and detection of pre-defined or new actions.
Monocular depth Estimation
Detect depth from single image for autonomous navigation & object avoidance.
Large Scale Object search
Search through millions of embeddings with in-GPU database, real-time.
Real-Time Embedded Solutions
All of our solutions have an embedded implementation for remote deployment.
Dense 3D Reconstruction
Dense 3D reconstruction with VR deployment for buildings, landscapes, and facilities.
Realtime Sparse 3D Reconstruction
Extremely accurate sparse 3D reconstruction for fast and reliable 3D map creation.
Zero Shot Detection
Learn to detect and track new objects without prior-training for immediate deployment.
Model R - driven by AI
Powerful & Compact
Integrated DSPs, GPUs, and SIMDs, 4K resolution; compact and impactful; built for real-time performance.
4G integrated. It can work at 3G speeds as well.
Learn more
MODEL R - [MR82N4]Modular, Scalable & Compact
In legacy camera systems, a captured video is encoded, streamed and then stored. Any video analytics is post processing. This makes handling 1000s of camera cluster installations, like those in Airports, Borders, Roads, etc. a nightmare. Loss of realtime crucial analytics is invaluable and post-processing of all stored streams simultaneously is impractical.
We have a nodal AI-Camera solution which captures the video feed, performs real-time video AI analytics, compresses the content, and then streams analyzed data, along with the video, for storage. It is a modern scalable solution designed for large-scale installations. No Fiber Optics required, as the system streams even on 3G connection. Compression also allows us to store 1YR video content in 1TB.
For more details, contact us.
INKERS
FOR ULTRA LARGE SCALE AIWhy choose Us.
We believe humans are explorers, adventurers, and creators, and AI must work on anything which stops us from being so.
Our Mission.
Design AI for ultra-large scale problems to keep humans productive, safe and healthy.
What we Do.
We repurpose our AI solutions for large enterprises to deliver large scale AI solution.
