KAVTAN

-A +A
KAVTAN Project is developed between 2008 and 2011. The main aim of KAVTAN Project is to improve RTÜK SKAAS System with the capability of automatic audio-visual concept recognition.
 
 

 

Concept is an abstract and generic design of an object or thought in human mind. Concept recognition is the process of teaching concepts to computers and enabling comptuters to detect them automatically.

The main target of KAVTAN Project is to evaluate the large scale of audio-visual data available in RTÜK SKAAS System. The data is automatically analyzed and certain concepts are detected, which is otherwise not feasible manually.

A hierarchic concept ontology is formed. The ontology is composed of five high level concepts and related low level concepts.

High Level Concept List

•    Violence
•    Nudity
•    Illegal Organization
•    Human Existence
•    Nature

Low Level Concept List

•    Audio Keyword
•    Videotext
•    Explosion Sound
•    Explosion 
•    Bullet Sound
•    Blood
•    Fire
•    Scream
•    Crying
•    Skin
•    Human Voice (male/female) 
•    Human Face
•    Face  
•    Crowd 
•    Cloud
•    Grass
•    Earth
•    Horizon
•    Artificial Edge
•    Speech
•    Music
•    Silence