A game-changer for computer vision training
Unity Computer Vision solutions help you overcome the barriers of real-world data generation by creating labeled synthetic data at scale. This data can be used to train computer vision models for object detection, image segmentation, and classification across retail, manufacturing, security, agriculture and healthcare.
Use cases
Retail
Grab and go: Detect shelf or cart items
Inventory: Audit shelves and ensure safety
Visual search: Deploy smart camera image recognition
Find threats: Identify abnormal store activity
Security
Crowd analysis: Track crowd movement and changes in pattern
Hazard monitoring: Detect intrusions or public safety threats
Manufacturing
Inventory management: Monitor warehouse inventory levels
Defect reduction: Identify abnormalities and issue alerts
Sorting: Reduce assembly line categorization issues
Agriculture
Plant health detection: Determine harvest time and analyze crop yield
Livestock management: Count, recognize and monitor livestock
Forestry: Monitor tree health and fire hazards
Key benefits
- Auto-labeled: No human annotation required
- Private: Compliant with privacy standards
- Safe: Recreate edge-case scenarios
- Iterative: Generate variations in datasets with simple code changes
- Variable and scalable: Produce training data that captures real-world complexity
- Affordable and accessible: Small ML teams can generate massive datasets within budget
Object-model training pipeline with synthetic data
There are three distinct stages in the synthetic data training pipeline:
- Content creation
- Synthetic data generation and analytics
- Model training
Content creation
There are several ways to create content. You can check out the Asset Store, scan your assets into Unity, or contact our team to help you with asset creation in Unity.
Synthetic data generation
Unity offers tools to generate synthetic datasets for use in perception-based computer vision tasks such as object detection, semantic segmentation, and more. You do not need prior experience with Unity or C# to get started.
Model training
You can either use your existing ML pipeline or implement using our suggested solution, Google AI Platform.
Blogs and videos
Can you find Waldo using synthetic data?
See how Unity’s Perception package was deployed to create Waldo-like images for training a neural net, which was then trained using the fastai library.
Create synthetic images for deep learning
Follow this tutorial to learn how to set up Unity and the Unity Perception package to create synthetic images that train neural nets in deep learning, AI and computer vision.
Myriad use cases enabled by synthetic data
Read how synthetic data is helping many organizations to overcome the challenge of acquiring labeled data for training machine learning models, and discover the breadth of use cases it enables.
Generate and analyze synthetic data at scale
Learn how to use tools from Unity to generate and analyze synthetic datasets with an illustrative example of object detection.
Train object detection model with synthetic data
Discover how you can generate a massive synthetic dataset to train your machine learning models.
Synthetic aided computer vision algorithm development
See how Standard Cognition used Unity to reduce the financial costs and algorithm development time for data collection and labelling in their digital checkout system.