What is AutoML Video Intelligence?
AutoML Video Intelligence enables training custom models for video analysis without machine learning expertise. The service supports video classification for categorizing entire videos and object tracking for following objects across frames. You provide labeled videos, Google trains an optimized model.
Core Features
- Video Classification: Categorize videos into custom classes
- Object Tracking: Detect and track objects across multiple frames
- Action Recognition: Identify actions and movement patterns
- Integrated Labeling: Web interface for efficient video labeling
- Batch Processing: Analysis of large video archives
Typical Use Cases
Manufacturing Quality Control
Cameras monitor production lines and detect defects in real-time. The model learns from examples of defect-free and defective products.
Sports Analysis
Tracking players and ball in sports videos. The model recognizes plays and automatically generates statistics.
Security Monitoring
Detection of unusual activities in surveillance videos. The model is trained on normal behavior and detects deviations.
Benefits
- Custom models for specific domains
- No deep learning knowledge required
- Scalable batch processing of large video archives
- Integration with Cloud Storage and BigQuery
Integration with innFactory
As a Google Cloud Partner, innFactory supports you with AutoML Video Intelligence: data preparation, labeling strategy, model training and integration into production systems. We help evaluate whether pre-trained APIs or custom models are a better fit.
Available Tiers & Options
AutoML
- No ML knowledge required
- Automatic training
- Fast labeling interface
- High training data requirements
Typical Use Cases
Technical Specifications
Frequently Asked Questions
What is AutoML Video Intelligence?
AutoML Video Intelligence enables training custom models for video analysis without ML expertise. The service supports video classification, object tracking and action recognition.
How many videos are needed for training?
For video classification, at least 100 videos per category are recommended. For object tracking, at least 1,000 annotated frames per object type should be provided.
What is the difference from the Video Intelligence API?
The Video Intelligence API offers pre-trained models for general video analysis. AutoML Video Intelligence trains custom models on your specific data for domain-specific tasks.
Is AutoML Video Intelligence part of Vertex AI?
Yes, AutoML Video Intelligence is integrated into Vertex AI. Training and deployment happens through the Vertex AI platform.
