Reference

AI for video engineering glossary

150 terms with crisp definitions, aliases, and cross-references to the deep dives where each is unpacked. From YOLO and SAM 2 to VLMs, LangGraph, and diffusion — every model, tool, and primitive in the AI for Video Engineering course in one place. Search by name, filter by phase, or jump straight to a letter.

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
ALL
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.

No terms match your filters. Try a different keyword or clear filters.

CrewAI

An agent framework that organises multiple role-based agents into a 'crew' that collaborates on a task. Popular for multi-agent workflows.

C

DeepFilterNet

A higher-quality open noise-suppression model that cleans speech well while staying efficient enough for real-time use.

D

DeepSORT

A widely used tracker that combines motion prediction with an appearance fingerprint to keep IDs stable through brief occlusions.

D

Depth Anything

A strong, general monocular depth model that estimates distance from a single image across varied scenes, small enough to run on-device.

D

Depth estimation

Predicting how far each pixel is from the camera, from ordinary 2D video. Enables 3D effects, occlusion-aware AR, and scene understanding.

D



Diffusion model

The dominant approach to generating images and video: start from noise and refine it step by step into a clean result guided by a prompt.

D

Echo cancellation

Stopping a speaker's own audio, played from the other side, from looping back into their microphone as an echo. Essential for hands-free calls.

E



Edge AI

Running models on or near the device that captures video — a camera, phone, or on-site box — instead of a distant data centre, to cut latency and bandwidth.

E

ElevenLabs

A leading commercial text-to-speech and voice-cloning provider known for natural voices, used widely for dubbing, narration, and conversational agents.

E



Embedding

A list of numbers that captures the meaning of an item — a word, image, face, or clip — so similar things sit close together and can be compared or searched.

E



EU AI Act

The European Union's comprehensive AI law, which classifies systems by risk and sets duties — including bans on some uses and transparency rules for others.

E

EU AI Act Article 50

The transparency rule in the EU AI Act requiring that AI-generated or manipulated media be disclosed to users. The reason synthetic content needs a label.

E

Face detection

Locating faces in an image with boxes — without identifying who they are. The first step before blurring, framing, or recognition.

F

Face recognition

Matching a detected face to a known identity. Powerful but legally sensitive — biometric rules and the EU AI Act constrain its use heavily.

F

Few-shot

Guiding a model with a handful of examples placed in the prompt, so it follows the pattern without any retraining. A quick way to steer behaviour.

F



Fine-tuning

Taking a general pre-trained model and training it a little more on your own data so it fits a narrow task without starting from zero.

F

FPS

Frames Per Second — how many images play or are processed each second. Higher FPS means smoother video but more frames to analyse and pay for.

F

Frame

A single still image in a video. Video is a rapid sequence of frames; most CV runs per frame, so frame count drives both quality and cost.

F

Frame sampling

Choosing which video frames to feed a model rather than all of them, to cut cost and tokens while keeping enough information to understand the clip.

F

Gemini

Google's family of frontier multimodal models, able to take text, images, audio, and video. Reached via API as a closed model, with very large context.

G



Generative AI

AI that creates new content — images, video, audio, text — rather than only analysing existing input. The basis of synthetic media features.

G

GGUF

A single-file format for distributing quantized language and multimodal models, popular for running them efficiently on CPUs and consumer GPUs.

G

GPT

OpenAI's line of frontier language and multimodal models, reached through a paid API. A default closed-model choice for reasoning and vision tasks.

G

GPU

Graphics Processing Unit — the parallel chip that runs almost all modern AI. GPU availability and price dominate the cost of self-hosting models.

G

Grounding DINO

An open-vocabulary detector that finds objects described by free-text prompts, not a fixed label list — detect 'a red backpack' without retraining.

G



Hallucination

When a model states something fluent but false or invented. The central reliability risk in language and multimodal AI, managed not eliminated.

H

Image classification

Assigning one label to a whole image — 'cat', 'weapon', 'safe' — without locating it. The simplest CV task and a building block for detectors.

I



Inference

Running a trained AI model on new input to get a prediction. The everyday cost of an AI feature comes from inference, not training.

I

Insertable Streams

A browser API that lets code modify each audio or video frame as it passes through WebRTC, the hook for in-call AI effects like blur and overlays.

I

Instance segmentation

Pixel masks that also separate individual objects — not just 'person pixels' but 'person 1' vs 'person 2'. Combines detection with segmentation.

I

Specialist software house for video, real-time and AI products. Founded 2005. 50 in-house engineers.

Knowledge base
BlogGuidesCoursesGlossaryDownloads
Company
ServicesProjectsDemosCalculatorContacts
+971-58-539-6367
WhatsApp
+1 (914) 775-5855
New York · USA
eager2develop@forasoft.com
Email
© Fora Soft, 2005–2026
Privacy PolicyTrust
Start the conversation
Describe your project and we will get in touch
Within 24 hours you get a realistic estimate, technical recommendations and next steps. Free, no obligations.
01
A senior architect reads it
Reply the same business day, Mon–Fri.
02
30-min call to scope it
We ask the questions that change the budget.
03
Estimate in 24 hours
Budget, timeline, architecture notes. Yours to keep.
Book a call
Pick a 30-min slot
WhatsApp us
Chat right now
@
Email us
eager2develop@forasoft.com
Or write here
I need
Enter your message
Enter your email
Enter your name

By sending this form you agree with the Personal Data Processing Policy.

Your message has been sent successfully
We will contact you soon
Message not sent. Please try again.