Available in VPC
Media Intelligence offers scene-level video analysis and automatic metadata extraction for videos and images. This allows you to search videos and images more easily and enhances indexing efficiency. While you can explore the service in detail with our Getting Started and Using Media Intelligence guides, we recommend starting with this quickstart overview to understand the complete workflow.
1. Upload files to Object Storage for analysis
Upload the files you want to analyze to the Object Storage bucket integrated with Media Intelligence. For detailed instructions, see:
2. Access workspace
Access the Media Intelligence workspace, where all tasks are performed. For detailed instructions, see:
3. Create project
Create a project, a space to manage media assets and analysis results. For detailed instructions, see:
4. Add media asset
After uploading files to an Object Storage bucket, add them to your Media Intelligence project. Analysis is performed based on the uploaded original files, and the video or image files registered in a project become the units of analysis and management, referred to as media assets.
For detailed instructions, see:
5. Media analysis
Analyze media assets using the multimodal engine (Sinossi) and extract metadata. Metadata is generated automatically based on the analysis output. For videos, this includes main keywords, scene summaries, and person information; for images, it includes key people, objects, and actions. You can analyze a single same media asset multiple times and generate various indexes based on the analysis settings. For detailed instructions, see:
6. View analysis results
On a separate webpage, you can view keywords, people, and caption information for each scene for videos, or the analysis results for images. This allows you to quickly find the edit points or search targets you need. You can also export the analysis results as a JSON file and use them for various purposes. For detailed instructions, see: