Forensics Investigation · POC #1
AI-Powered CCTV Forensics Investigation
Identify key moments, retrieve traces in seconds. Full suspect profiling, event reconstruction, accomplice behavior analysis, and court-ready evidence report — generated from raw CCTV footage. No manual tagging. No analyst hours.
🚨 Threat Level: Critical
CRITICAL: Coordinated Shoplifting Incident Detected — Two Suspects
- 00:00:05 (CCTV: 00:01:23) — P1 conceals item O1 into personal shoulder bag
- 00:00:05 (CCTV: 00:01:23) — P2 makes direct eye contact with P1 during concealment — coordination confirmed
- 00:00:17 (CCTV: 00:01:35) — Both suspects exit visible frame together without proceeding to checkout
Evidence Quality Score
8.8/ 10
🚨 CRITICAL Threat2 Persons of Interest3 Flagged EventsExcellent · A Grade
This CCTV footage provides excellent evidential support for a coordinated shoplifting incident. Clear visualization of merchandise concealment and suspicious accomplice behavior allows for robust investigative and prosecutorial efforts. Evidence quality rated 88% of ideal forensic standard.
Forensic Metrics · 14 Detected
Person Tracking Accuracy
100%
Both P1 and P2 tracked continuously from 00:00:00 to 00:00:20 without identity loss.
Suspicious Activity Classification
100%
Concealment, counter-surveillance, coordination signal, and checkout evasion all classified.
Timestamp Extraction Accuracy
100%
CCTV display timestamps (00:01:18:08) precisely correlated to video playback time.
Entry/Exit Event Detection
100%
Exit event at 00:00:17 detected; implied entry before 00:00:00 noted.
Clothing Description Accuracy
90%
P1: blue jacket, polka-dot skirt, Adidas sneakers. P2: beige jacket, stripes, round glasses.
Movement Pattern Analysis
95%
Full path reconstruction for both suspects from entry to evasion exit.
Height/Body Size Estimation
90%
P1: 160–165 cm. P2: 170–175 cm. Estimated relative to shelf height.
Interaction Event Detection
85%
P2 critical glance at P1 (00:00:05) and coordinated exit detected. No audio for verbal comms.
Object Identification Quality
70%
Small rectangular package (O1) identified. Exact product brand not determinable at camera distance.
Facial Recognition Quality
Medium
Overhead angle and obstructions limit high-confidence FR. Usable with enhancement.
Video Quality Score
9/10
High resolution, excellent clarity, consistent bright lighting, stable camera throughout.
Chronological Event Sequencing
100%
All 12 events documented in precise chronological order with CCTV timestamps.
Persons of Interest · Suspect Profiles
P1
Person 1 — Primary Actor
Shoplifting · Merchandise Concealment
Female, mid-20s to early 30s. Blue long-sleeve jacket over dark inner top. Dark polka-dot mid-calf skirt. Adidas sneakers (3 black stripes). Black shoulder bag used for concealment. Dark curly/wavy hair, shoulder length.
Height
160–165 cm
Build
Average–Slim
Age Estimate
Mid-20s–30s
Threat Level
CRITICAL
00:00:05 — Conceals item O1 into shoulder bag (THEFT)
00:00:06 — Counter-surveillance scan of aisle
00:00:02 — Glances toward P2 before concealment
00:00:17 — Exits without approaching checkout
P2
Person 2 — Possible associate (AI-suggested)
Observed: lookout-consistent positioning · coordination · distraction
Male, mid-20s to early 30s. Beige/light brown zip-up jacket over horizontally striped top (alternating dark/light). Dark trousers. Dark athletic shoes. Round glasses. Dark curly hair. Dark backpack worn on back.
Height
170–175 cm
Build
Average
Age Estimate
Mid-20s–30s
Threat Level
MEDIUM-HIGH
00:00:05 — Direct glance at P1 during concealment (coordination signal)
00:00:06 — Resumes casual browsing as cover
00:00:11 — Leads P1 toward exit
00:00:17 — Joint exit without checkout — accomplice confirmed
Incident Timeline · 12 Events · 17 Second Span
00:00:00
Monitoring begins — P1 & P2 in aisle
Both persons already present in snack/candy aisle. P1 near lower shelf, P2 at higher shelf further down.
P1P2
CCTV: 00:01:18:08
00:00:01
P1 picks item O1 from bottom shelf
P1 reaches down with right hand and picks up a small rectangular item from the lowest shelf.
P1
CCTV: 00:01:19:08
00:00:02
P1 brings item toward body · glances at P2
P1 brings O1 closer to her body and briefly glances left toward P2 — possible pre-theft check.
P1
CCTV: 00:01:20:08
00:00:03
P1 opens shoulder bag with left hand
Left hand moves deliberately to open the black shoulder bag in preparation for concealment.
P1
CCTV: 00:01:21:08
00:00:05
Criminal Activity
🚨 P1 CONCEALS item O1 in shoulder bag
P1 places item O1 directly into her open black shoulder bag with her right hand — clear act of theft.
P1
CCTV: 00:01:23:08
00:00:05
Suspicious Activity
P2 glances at P1 — coordination signal
At the exact moment of concealment, P2 makes a quick direct glance toward P1 — strongest indicator of coordination.
P2
CCTV: 00:01:23:08
00:00:06
Counter-Surveillance
P1 scans aisle after concealment
P1 quickly closes bag and looks around the aisle — classic counter-surveillance behavior post-theft.
P1
CCTV: 00:01:24:08
00:00:06
P2 resumes casual browsing as cover
P2 immediately returns to appearing to browse merchandise — providing distraction/cover after coordination glance.
P2
CCTV: 00:01:24:08
00:00:11
P1 moves to join P2
P1 walks closer to P2, who has shifted position toward the exit path — coordinated withdrawal begins.
P1P2
CCTV: 00:01:29:08
00:00:16
P2 leads exit — P1 follows
P2 moves toward upper-right exit of frame, P1 follows. No deviation toward checkout area.
P1P2
CCTV: 00:01:34:08
00:00:17
Criminal Activity
🚨 Both suspects exit frame — checkout evasion confirmed
P1 and P2 exit the camera's visible area together, following a path that bypasses all checkout counters. Theft confirmed.
P1P2
CCTV: 00:01:35:08
00:00:17
Alert
⚠ ALERT: Coordinated Shoplifting Confirmed
Incident classification: Coordinated Shoplifting. Investigation Priority: URGENT. Alert law enforcement immediately.
P1P2
CCTV: 00:01:35:08
Evidence Quality Assessment
Video Quality Score
9 / 10
High resolution, excellent clarity, stable camera, consistent lighting throughout 20-second segment.
Person ID Potential
High
Clothing details (Adidas brand, skirt pattern, glasses) are highly distinctive and identifiable.
Facial Recognition Quality
Medium
Overhead angle limits FR. Usable for comparison with enhancement. Best frames at 00:00:05 and 00:00:07.
Chain of Custody
Not established
AI analysis is an investigative aid — custody must be established through proper forensic extraction of the original footage.
Object Identification
Low–Med
Item O1 presence confirmed. Specific product type/value not determinable from camera distance.
Report Completeness
100%
Full forensic report generated from raw CCTV upload — suspect profiles, event timeline, BOLO report, and evidence quality assessment.
Recommended Immediate Actions
CRITICALIssue BOLO alert to local law enforcement with detailed descriptions of P1 (blue jacket, Adidas shoes, black shoulder bag) and P2 (beige jacket, round glasses, backpack). Incident type: coordinated shoplifting.
CRITICALImmediately secure all CCTV footage from entry/exit cameras and parking areas to track suspects' full movement and identify any vehicle/license plate.
HIGHConduct inventory check in the snack/candy aisle for losses of small rectangular items. Review POS data for CCTV timestamp 00:01:18–00:01:38 to confirm no purchase was made.
HIGHExtract high-resolution still frames from 00:00:05 (P1 concealment) and 00:00:07 (P1 face upward glance) and 00:00:05 (P2 profile) for BOLO distribution and FR submission.
HIGHDisseminate suspect descriptions and modus operandi (coordinated entry, aisle positioning, bag concealment, cover-and-exit) to regional loss prevention network.
MEDIUMCross-reference suspect clothing descriptions against any local CCTV databases, known-offender systems, or retail loss prevention watchlists.
Ideal for Retail Security, Loss Prevention & Law Enforcement
VidForgeX Forensics gives security teams, loss prevention departments, and law enforcement agencies the same depth of analysis a full forensics unit would provide — automated, standardized, and delivered in minutes from any CCTV upload.
[email protected]Mohamed Fazaary · RaynX.ai
Frequently Asked Questions
How fast does VidForgeX analyse CCTV footage?
VidForgeX processes CCTV footage rapidly — short clips (under 1 minute) typically complete within minutes, including full suspect profiling, event timeline reconstruction, accomplice detection, and a BOLO-ready evidence report. Longer clips (10–30 minutes) typically complete in 3–6 minutes.
What types of CCTV footage does VidForgeX support?
VidForgeX supports any standard CCTV export format: MP4, MOV, AVI, and MKV. It works with overhead angles, side angles, wide-angle store cameras, and dashcams. Resolution from 480p up to 4K is supported, though 720p+ yields the best descriptor accuracy.
Can VidForgeX detect coordinated criminal behavior?
Yes — this is a core capability. VidForgeX analyzes inter-person behavioral correlation: synchronized glances, timing of movements relative to each other, division of roles (actor vs. lookout), and joint departure patterns. In this POC, P2's lookout role was identified from a single-frame glance at the exact moment P1 concealed merchandise.
What does the suspect profile include?
Each suspect profile includes: height estimate (±5 cm), build, age range, hair description, clothing breakdown (brand, color, pattern, style), distinctive accessories, facial recognition quality score, best-frame timestamp for FR extraction, gait pattern notes, and a full timestamped action log of every observed behavior.
Is the output court-ready and chain-of-custody safe?
VidForgeX output is an investigative aid, not certified forensic evidence — it does not by itself satisfy chain-of-custody requirements. Each event is tied to a CCTV display timestamp (e.g., 00:01:23:08), and reports include an evidence quality assessment and preservation notes so your team can pursue proper forensic extraction of the underlying footage. AI-suggested groupings are labelled as such and remain distinct from investigator conclusions.
How do I get my CCTV footage analysed by VidForgeX?
You can upload footage directly at forensic.vidforgex.com or contact Mohamed Fazaary at [email protected] to discuss a custom integration for your security operations, loss prevention team, or law enforcement agency.