About Spectre
A proactive intelligence assistant for Indian policing.
Spectre was built for the Karnataka State Police × Hack2Skill KSP Datathon 2026 — Challenge 2: AI-Driven Crime Analytics & Visualization Platform. Our mandate: turn fragmented FIRs and historical case data into an immersive, decision-grade command surface that officers can act on inside one shift.
Karnataka — 5 year snapshot
Over 1.04 million cognizable crimes registered in 5 years.
Source: NCRB Crime in India 2019-2023 (cognizable IPC + SLL, Karnataka)
2019
194,118
Pre-pandemic baseline year for cognizable crimes in Karnataka.
2020
169,014
Sharp dip during lockdown; cyber complaints rose against the trend.
2021
199,929
Recovery year, with cyber and economic offences accelerating.
2022
240,222
Post-pandemic peak; property and cyber crimes both spiked.
2023
237,228
High plateau — cyber crime now an outsized share of new FIRs.
≈1.04M
FIRs across 5 years
+22%
Surge from 2020 → 2022
2-3x
Cyber-crime growth since 2019
Even with capable district teams, the sheer volume — over 600 new cognizable cases every day on average — means patterns get noticed late. Spectre compresses that signal-to-decision gap from days into minutes.
How Spectre solves it
Four pillars, one operations advantage.
Predictive hotspots
Frequency, week-on-week trend and concentration combine into a single 0-100 risk score per zone.
Plain-language briefings
AI explains WHAT changed, WHY it likely happened, WHAT NEXT, and a CONFIDENCE band — never a black box.
Patrol optimisation
Recommended day, time window and patrol type per zone, with print-ready briefings for the SHO.
Real-time alerts
Spike, time-window, category surge, new hotspot and repeat-location detectors push SMS to officers via gateway.
Principles
A decision-support tool, never a replacement.
Spectre augments — it does not replace — the judgement of the officer on the ground. Every recommendation is auditable; every score is explained.