Media Data Processing
Acquisition and conversion of video and audio data
Acquisition and conversion of video and audio data
Media data is increasingly leveraged in every industrial situation. Work is automated based on analysis of video and audio, and image recognition AI improves the work efficiency. Now, media data including video and audio is essential to machine learning, the development of AI, and other advancements. To develop an AI model that makes the most of media data (e.g., computer vision and audio analysis) or to run a continuous learning cycle of a model, you must build a complicated pipeline that includes data acquisition and conversion to collect media data, make your machine continuously learn models, and deploy new models in automation operation. The intdash-based media data pipeline comprehensively provides mixed functionality required for such digital transformation, including transmission, storage, and conversion of video and audio.
Use cases
Acquisition of learning data and validation data in AI development
Acquisition of video and audio data for event analysis
Preprocessing to prepare learning data
Real-time video streaming for remote control
Challenges related to acquisition of video and audio data
Acquisition of media data such as video and audio has many different challenges including support for various codecs, timestamp matching with other fusion data, and transcoding after collection.
- The receiving client may not support the codec that the sending client uses.
- The high efficiency video codec appropriate for transmission may not match the image codec suitable for machine learning.
- Data zapping and integrated management of massive media data and other data
- Timestamp synchronization between media data and other sensor data
Main features of DX Functions
Many-to-many video live stream
Real-time many-to-many streaming of video data via a server
View DetailsUse of media data for machine learning
Conversion of high-compression video data to an image data format that is easy to use in the annotation process of machine learning
View DetailsIntegrated timestamp processing for media data
Adding integrated timestamps to data streams across multiple sources such as "video and video" or "video and sensor signal" to map data to the same time axis
View DetailsLow-latency video streaming
Real-time conversion of video data into streaming formats such as HLS and WebSocket
View DetailsSupport for audio data
Various types of audio data including AAC and PCM are supported. Audio data is transmitted and saved as time-series data.
View DetailsMedia data processing pipeline
You can build a data pipeline to process collected raw media data for secondary use.
View DetailsMEDIA DATA PROCESSING system reference
