Skip to content

Video Extractors

Video extractors are specialized tools designed to process and analyze video content, extracting valuable information from the visual and audio components of video files. These extractors can perform various tasks such as frame extraction, audio separation, face detection, and object tracking. By breaking down videos into their constituent parts or identifying specific elements within them, video extractors enable advanced analysis, indexing, and content understanding for a wide range of applications in media processing, surveillance, and content management systems. If you want to learn more about extractors, their design and usage, read the Indexify documentation.

Extractor Name Use Case Supported Input Types
Key Frame Extractor Extract key frames from videos video, video/mp4
Audio Extractor Extract audio from video files video, video/mp4, video/mov, video/avi
Face Extractor Extract and cluster unique faces from videos video, video/mp4
Track Extractor Track objects in videos using YOLO video, video/mp4

Key Frame Extractor

Description

Dissect a video into its key frames and generate JPEG image content for each extracted frame. Comes with frame frequency and scene detection configuration that can be changed with input parameters.

Input Data Types

["video", "video/mp4"]

Class Name

KeyFrameExtractor

Download Command

indexify-extractor download tensorlake/scene-frame-extractor

Audio Extractor

Description

Extract the audio from a video file.

Input Data Types

["video", "video/mp4", "video/mov", "video/avi"]

Class Name

AudioExtractor

Download Command

indexify-extractor download tensorlake/audio-extractor

Face Extractor

Description

Extract unique faces from video. This extractor uses face_detection to locate and extract facial features and sklearn DBSCAN to cluster and find uniqueness.

Input Data Types

["video", "video/mp4"]

Class Name

FaceExtractor

Download Command

indexify-extractor download tensorlake/face-extractor

Track Extractor

Description

A YOLO-based object tracker for video.

Input Data Types

["video", "video/mp4"]

Class Name

TrackExtractor

Download Command

indexify-extractor download tensorlake/tracking