AI-Based Intelligent Video Analytics Platform for Border Surveillance using existing CCTV Infrastructure.
Ministry of Home Affairs · Smart Automation · Software
The no-extra-hardware value is real and border-relevant, but this is a saturated category with a long feature list — build the intrusion-detection and ANPR core well on real CCTV footage and be honest about the vague suspicious-activity part rather than claiming the whole list.
What it actually is
Border forces have lots of ordinary CCTV cameras but they only record and need constant human watching, while smart features like face recognition, plate reading and intrusion detection usually need expensive proprietary hardware. The ask is a software platform that turns existing standard CCTV into an intelligent surveillance network — running detection, tracking, ANPR, intrusion alerts and more purely in software.
What to build
A software-defined video-analytics platform that ingests live streams from standard IP CCTV cameras and runs, in software, human detection and tracking, vehicle detection and classification, face detection, automatic number-plate recognition, virtual-fence intrusion detection, suspicious-activity and night-time-movement detection, with real-time alert generation and event logging, so existing border CCTV becomes an intelligent network without dedicated FRS, ANPR or smart-camera hardware.
Smallest thing that wins the room
Point the platform at a standard CCTV stream and show it drawing a virtual fence, detecting and tracking a person crossing it at night, classifying a vehicle and reading its plate, and firing a real-time intrusion alert with the event logged.
How crowded this one gets
A guess, projected from the 2025 statements — the last year where both the submission counts and the winners were published.
Quieter than 61% of the 226 · #88 of 226 by expected field
A normal-sized field. Your idea has to be good, not miraculous.
Why: defence, intelligence and space bodies drew small fields.
This is a guess, not a fact
Nobody has published 2026’s numbers yet. This is an analysed estimate from last year’s pattern, so please do not take it as the truth — check the live counter on the SIH portal before you decide anything. The range covers the middle half of likely outcomes, so one statement in two lands outside it. Entry closes at 500 ideas per statement, so no range goes past that — a statement that reaches the cap fills and shuts rather than drawing an unlimited crowd. The model reads only three things a team can see before choosing — software or hardware, the theme, and what kind of body posted it — and those explain about a quarter of the variation in last year’s field sizes (R² 0.25 on held-out statements). Trust the band more than the number, and the ordering more than either. It cannot see how good your idea is, which is the part that actually decides it.
The scores
The number is the shorthand. The line under it is the reason.
Acceptance potential
3/5The no-dedicated-hardware value is genuine and border-relevant and the individual capabilities are achievable, but video analytics over CCTV is a saturated category, the feature list invites shallow breadth, and suspicious-activity detection is vague enough that a team will demo the easy detections while under-delivering the hard ones.
Feasibility
3/5Detection, tracking, ANPR and virtual-fence intrusion are all standard computer vision buildable on public datasets, but bundling all of them to run reliably in real time on many standard CCTV streams is a broad build, and suspicious-activity detection in particular is vague and hard to do well.
Innovation scope
2/5Software video analytics over CCTV is a mature commercial category, and the description prescribes the standard feature set, so beyond the run-on-existing-hardware angle there is little genuinely new.
Clarity
4/5The capabilities are enumerated clearly and the core value — no dedicated hardware — is explicit, though suspicious-activity detection is left undefined.
Effort
MassiveHuman and vehicle detection, tracking, face detection, ANPR, virtual-fence intrusion, activity and night detection, plus alerting and logging across live streams, is a broad multi-feature platform.
Demo-ability
EasyA person crossing a virtual fence at night triggering a live alert on real CCTV footage is immediately legible and directly on-mission.
In its favour
- Green flag: Detection, tracking, ANPR and virtual-fence intrusion are all standard and buildable on public datasets and real CCTV footage
- Green flag: The no-dedicated-hardware framing is a genuine, border-relevant value that reuses existing infrastructure
- Green flag: A virtual-fence intrusion alert at night is an immediately legible, on-mission demo
- Green flag: Running on standard IP streams keeps the deployment story realistic for remote border posts
Against it
- Red flag: Video analytics over CCTV is a saturated commercial category, so a competent build resembles existing products
- Red flag: The feature list is long, and running all of it reliably in real time on standard streams invites shallow breadth
- Red flag: Suspicious-activity detection is vague and genuinely hard, so it tends to be the feature that is claimed but not delivered
- Red flag: Face recognition at a border raises accuracy, bias and privacy questions the description does not address
What you will be writing
- YOLO detection + DeepSORT tracking on RTSP streams
- ANPR (plate detection + OCR)
- Face detection (RetinaFace)
- Virtual-fence intrusion logic
- Real-time alerting + event logging
- Night-time / low-light enhancement
- Video surveillance analytics
- Border security
- Computer vision
Prior art to read before you start
CCTV-to-smart-surveillance software · virtual-fence intrusion detection · multi-capability video analytics
Analysed by Claude Opus. Every score above is a judgment call with its reasoning attached — kindly cross-check this against the official statement on the SIH portal before your team commits to it.