§ 06  /  The Glossary

A vocabulary for computer vision in security & operations.

152 plain-English definitions — cameras and optics, networking, edge AI, detection and analytics, video search, and India's DPDP regime. Search it, or jump by letter.

152 / 152 terms
No terms match that search.
#
3D DNR

Three-dimensional digital noise reduction for cleaner low-light video.

A
Abandoned Object

Alerting on a bag or item left unattended.

Alert Fatigue

Operators ignoring alerts because too many are false.

Annotation

Labelling objects in data to create training examples.

Anomaly Detection

Flagging behaviour that deviates from the norm.

Anonymisation

Irreversibly stripping data of identifying information.

ANPR

Automatic Number Plate Recognition — extracting license-plate text from video.

Appization

IndoAI's approach of packaging an AI capability as a programmable, deployable app that runs on the edge camera platform.

Audit Log

A tamper-evident record of who accessed or exported video.

B
Backlight Compensation (BLC)

Exposure adjustment so subjects aren't silhouetted by bright backgrounds.

Bandwidth

The network capacity, in Mbps, available to carry video.

Behaviour Analysis

Interpreting actions such as fighting or running.

Bitrate

Data per second used to encode the stream, trading quality against bandwidth and storage.

Bounding Box

The rectangle a detector draws around an object.

Bullet Camera

A cylindrical, outward-facing camera, common outdoors.

C
Classification

Assigning a category label to an image or detected object.

CNN

Convolutional Neural Network — the backbone of most vision models.

Codec

The algorithm that compresses (encodes) and decompresses (decodes) video.

Confidence Score

The probability a model assigns to a detection being correct.

Crowd Density

Estimating how many people occupy an area.

D
Data Fiduciary

The entity that decides why and how personal data is processed.

Data Minimisation

Collecting and keeping only the data actually needed.

Data Principal

The individual the personal data relates to.

Dataset

The collection of images or video used to train or evaluate a model.

Day/Night Camera

A camera that switches between colour and IR modes by light level.

DHCP

Automatic IP-address assignment on a network.

Distillation

Training a small model to mimic a larger one.

Dome Camera

A ceiling-mounted camera in a dome housing, common indoors.

DPDP Act

India's Digital Personal Data Protection Act, 2023, governing the processing of personal data, including identifiable video.

DVR

Digital Video Recorder — an older recorder for analog cameras over coax.

Dwell Time

Time a person spends in a defined zone — a retail KPI.

E
Edge AI

Running inference on or near the camera instead of in the cloud.

Edge AI Box (EdgeBox)

An on-site appliance that runs AI inference on existing camera feeds.

Edge Storage

Recording onto a card or disk inside the camera or a local appliance.

Embedding

A numeric vector representing an image, face, or object for comparison.

Event Clip

A short recording bookmarked around a triggered event.

F
F1 Score

The harmonic mean of precision and recall.

Face Detection

Locating faces in a frame, without identifying them.

Face Recognition

Matching a detected face to a known identity.

Facial Landmarks

Key points on a face used for alignment or analysis.

Failover

Automatic switch to backup recording when a device or link fails.

Fall Detection

Recognising when a person has fallen.

False Alarm

An alert triggered by something that wasn't a real event.

Field of View (FOV)

The angular extent of a scene a camera captures, set by lens focal length and sensor size.

Fine-Tuning

Adapting a pre-trained model to a specific use case or site.

Fire & Smoke Detection

Spotting flames or smoke in video at an early stage.

Fisheye Camera

An ultra-wide 180–360° camera that is de-warped in software.

Fixed Lens

A lens with a single, non-adjustable focal length.

FNR

False Negative Rate — how often a system misses a real event.

Focal Length

The lens property, in millimetres, that determines how wide or zoomed the view is.

Footfall

The count of people entering a space over time.

FPR

False Positive Rate — how often a system raises an unnecessary alarm.

Frame Rate (FPS)

Frames captured per second; affects motion smoothness and bandwidth.

G
GOP

Group of Pictures — the spacing between full keyframes in a compressed stream.

GPU

A graphics processor often used to accelerate AI inference.

Ground Truth

The verified correct labels a model is measured against.

H
H.264 (AVC)

A widely used video-compression standard.

H.265 (HEVC)

A newer codec that roughly halves bitrate versus H.264 at similar quality.

Heatmap

A visual map of where movement or dwell concentrates over time.

Helmet Detection

Checking for hard hats or two-wheeler helmets.

Hybrid Architecture

Splitting workloads between edge devices and the cloud.

Hygiene Compliance

Checking hand-wash or cleaning routines via video.

I
I-Frame

A fully self-contained keyframe used as a compression reference point.

IK Rating

Impact-protection code indicating vandal resistance.

Image Sensor

The chip (CMOS or CCD) that converts light into the video signal.

Inference

Running a trained model on live data to produce predictions.

Intrusion Detection

Alerting when a person enters a restricted zone.

IoU

Intersection over Union — an overlap-based measure of detection accuracy.

IP Camera

A network camera that streams video over an IP network rather than coax.

IP Rating

Ingress-protection code (e.g. IP66/67) for dust and water resistance.

IR Illumination

Infrared LEDs that let a camera see in darkness, rendered as monochrome.

L
Latency

The delay between an event happening and it appearing or being acted on.

Line Crossing

An alert when an object crosses a defined virtual line.

Loitering Detection

Flagging a person lingering in an area beyond a threshold.

LPR

License Plate Recognition — the term used outside India for ANPR.

Lux

A measure of light level; low-lux ratings indicate low-light performance.

M
Mainstream

The high-resolution primary stream from a camera, used for recording.

mAP

mean Average Precision — a standard accuracy metric for detectors.

Mask Detection

Checking whether a face covering is worn.

Megapixel (MP)

One million pixels — a shorthand for camera resolution.

Metadata

Structured tags — objects, time, zone — attached to video for search.

MJPEG

A simple codec encoding each frame as a separate JPEG; high bandwidth.

Model

A trained neural network that performs a task such as detection.

MOT

Multi-Object Tracking — tracking many objects simultaneously.

Motion Detection

Basic pixel-change detection that triggers recording or alerts.

Motorised Lens

A varifocal lens whose zoom and focus are driven by motors for remote adjustment.

Multicast

Sending one video stream to many receivers efficiently over a network.

N
NMS

Non-Maximum Suppression — removing duplicate overlapping detections.

NPU

Neural Processing Unit — a chip specialised for AI inference at the edge.

NTP

Network Time Protocol — keeps camera and recorder clocks in sync for accurate timestamps.

NVR

Network Video Recorder — a device that records and manages IP-camera streams.

O
Object Detection

Locating and classifying objects in an image with bounding boxes.

Object Tracking

Following a detected object across frames.

On-Premise

Running systems on local hardware rather than in the cloud.

ONNX

An open format for exchanging trained models between frameworks.

ONVIF

An open standard letting CCTV cameras, recorders, and software interoperate across brands.

Overfitting

When a model memorises training data and fails on new data.

P
P-Frame

A frame stored only as the difference from previous frames.

People Counting

Tallying individuals crossing a line or occupying a zone.

Personal Data

Information that can identify an individual, such as a face or plate.

PoE

Power over Ethernet — delivering power and data to a camera over one network cable.

Pose Estimation

Locating body keypoints to infer posture or action.

PPE Compliance

Detecting whether workers wear required safety equipment.

PPM (Pixels per Metre)

How many horizontal pixels a camera resolves across one metre at a given distance — the core metric for whether it can detect, recognise, or identify.

Precision

Of all alerts raised, the share that were correct.

Privacy Mask

A fixed, blocked-out region a camera never records.

Pruning

Removing redundant weights to make a model smaller and faster.

Pseudonymisation

Replacing identifiers with tokens that can be reversed under control.

PTZ Camera

Pan-Tilt-Zoom camera that can be steered and zoomed remotely.

Q
Quantization

Shrinking a model (e.g. to INT8) so it runs faster on edge hardware.

Queue Analytics

Measuring queue length and wait time.

R
RAID

Disk redundancy that protects recordings against drive failure.

RBAC

Role-Based Access Control — limiting who can view or export footage.

Re-ID

Re-Identification — matching the same person or vehicle across non-overlapping cameras.

Recall

Of all real events, the share the system caught.

Redaction / Masking

Blurring faces or plates to protect identity.

Region of Interest (ROI)

A defined area where analytics are applied.

Resolution

The pixel dimensions of the video, e.g. 1080p, 4MP, or 8MP (4K).

Retention

How long recorded video is kept before being overwritten.

RTMP

Real-Time Messaging Protocol — a streaming protocol used mainly for ingest to servers.

RTSP

Real-Time Streaming Protocol — how IP cameras and NVRs ship live video.

S
Segmentation

Labelling an image pixel-by-pixel rather than with boxes.

Slip & Fall

Detection of slips, trips, and falls for safety or claims.

Smart Compression

Encoding important regions at higher quality to save storage.

Smart Parking

Detecting free and occupied parking bays.

Speed Estimation

Estimating vehicle speed from video.

SRT

Secure Reliable Transport — a low-latency, loss-tolerant streaming protocol.

Starlight

A label for sensors that keep colour images in very low light.

Static IP

A fixed, manually assigned network address for a device.

Substream

A lower-resolution secondary stream used for live preview or analytics.

T
Tailgating Detection

Spotting an unauthorised person following through a door.

TOPS

Tera-Operations Per Second — a rough measure of an accelerator's AI throughput.

Training

The process of teaching a model from labelled data.

Tripwire

A virtual line that triggers an event when crossed.

Turret Camera

An eyeball-style camera offering easy aiming and reduced IR glare.

V
Vandal-Resistant

A housing built to survive tampering or impact.

Varifocal Lens

A lens whose focal length (zoom) can be adjusted manually or by motor.

Vehicle Classification

Distinguishing cars, trucks, two-wheelers, and the like.

Video Summarization

Condensing hours of footage into a short reviewable clip.

Virtual Fence

A drawn boundary monitored for crossings or intrusion.

Vision Transformer (ViT)

A transformer-based architecture applied to images.

VLAN

A virtual LAN used to isolate camera traffic from other network traffic.

VMS

Video Management System — software that records, displays, and manages many cameras.

VSaaS

Video Surveillance as a Service — cloud-hosted recording and management.

W
WDR

Wide Dynamic Range — processing that balances very bright and dark areas in one frame.

WebRTC

A browser-native protocol for low-latency live video.

Wrong-Way Detection

Flagging vehicles moving against the traffic flow.

Y
YOLO

A family of real-time object-detection models (“You Only Look Once”).