Video intelligence that sees what it should — and nothing more.
Aeye Vision builds the infrastructure layer for privacy-preserving video AI. Analysis happens at the edge, identities stay encrypted, and every viewer sees only the people they are authorised to see.
3core video-intelligence patent applications filed in India
0raw frames leave the edge device by design
1stream, many viewers, each with their own key
Edge-first
Detection, pose and activity analysis run on the device next to the camera. What travels upstream is structured insight, not footage.
Abstain, don't guess
When a frame isn't good enough to judge, the system says so. No bare yes/no outputs; conclusions are built from evidence over time.
Compliant by architecture
Designed from the first line for India's DPDP Act 2023 and DPDP Rules 2025 — consent, purpose limitation and auditability are structural, not bolted on.
Intellectual property
Three patents, one privacy-preserving stack.
A foundation for privacy-preserving analysis, an application that turns it into developmental insight, and a cryptographic layer that decides who may see whom.
IN 202541112271Foundation
Privacy-Preserving Real-Time Video Intelligence System for Educational Developmental Analytics
An edge-based video intelligence system for learning environments. Children are understood through movement, activity and interaction rather than identity: faces are protected on-device before anything leaves the classroom, and observations from video, photos, activity records and educator notes are combined into a privacy-preserved developmental picture.
On-device anonymisation before transmission
Activity understanding without face identification
Consent and retention built into the architecture
IN 202541123399Application
Developmental Milestone Detection
A real-time, multi-modal system that recognises developmental milestones as they happen in everyday settings — play, meals, group activity — instead of in occasional assessments. Milestones are confirmed only when evidence is consistent over time, and possible delays are surfaced early for educators and specialists to review.
Continuous, naturalistic observation
Evidence accumulated over time before confirmation
Early flags for human review, not automated diagnosis
IN 202541112901Platform
Identity-Based Encryption for Personalised Selective Video Unmasking (IBE)
Each identity in a video is encrypted separately, so one stream can be shared with many viewers while each sees only the people their key allows. A parent sees their own child. An investigator with authorisation sees the people involved in an incident. Everyone else stays masked — cryptographically, not by policy promise.
Per-identity encryption in a single stream
Viewer-specific keys with scoped, auditable access
Applicable well beyond education
Descriptions summarise the inventions at a high level. Technical implementation details are available to partners under NDA.
How it fits together
From camera to insight, without exposing a face.
01
Capture
Existing CCTV or a low-cost AI vision sensor feeds an edge device on site.
02
Understand at the edge
People, posture and activity are analysed locally. Unusable frames are flagged, not guessed at.
03
Encrypt per identity
Each person in view is sealed under their own identity key before storage or transmission.
04
Aggregate over time
Observations become developmental trends and milestone evidence across days, not single frames.
05
Authorised view
Educators, parents and auditors each receive exactly the view their role and consent allow — every access logged.
Interactive illustration
Same stream. Different keys. Different views.
Pick a viewer. The underlying video never changes — only which identities that viewer's key can decrypt.
Without a key, every person in the frame stays masked.
CAM-03 · encrypted stream0 / 6 visible
Arjun
Meera
Kabir
Zoya
Staff
Visitor
Names are fictional. Illustration only.
Where it applies
Built for early childhood. Designed to travel.
Early childhood development
Anganwadi centres, corporate crèches and preschool chains get continuous developmental insight and child safety monitoring — with parents seeing only their own child.
Governance & public safety
Smart-city and institutional video where analytics run at scale while citizens stay masked unless lawful authorisation unlocks a specific identity.
Camera & VMS makers
Selective unmasking as an embedded capability for privacy-regulated deployments.
Healthcare
Monitoring where each clinician sees only the patients in their care.
Judicial & evidentiary video
Proceedings and evidence shared with witness and bystander identities protected.
Video collaboration
Recordings where each participant's visibility follows their access rights.
Licensing, pilots and partnerships.
We work with camera and silicon makers, early-childhood operators, and public-sector programmes. Technical deep-dives are available under NDA.
