Annotation Products
Pivotal tools for AI-driven computer vision and product labeling
AI-Driven Product Labeling
The Annotation Products section is pivotal for the AI computer vision system. By accurately labeling product images, this section ensures the AI can detect and link images to product information.
Key Capabilities
Annotation Importance
Critical for AI product detection accuracy
The Foundation of Recognition
Accurate image annotation is critical for the AI to reliably detect products during checkout and inventory processes. Labels connect images to product data, ensuring seamless integration with the AI model for optimal performance.
Annotation Features
Tools and capabilities for image labeling
Image Upload
Upload at least two images per product to assign to an AI model. A modal with a tagging interface allows precise labeling, supporting multiple nested labels within a primary label.
Image Scanning
Automatically capture images using configurable settings (delay, number of images, brightness). Labels are drawn and applied dynamically after scanning.
Labeling System
Cartify's annotation system links images to products via unique labels, enabling accurate AI detection. Nested labels must be contained within a primary label.
Frame-Based Detection
Use a frame to detect product size or distance by positioning the product within a designated area, enhancing detection accuracy for specific scenarios.
Uploading Images
To assign images to the AI model, upload at least two images per product. Click the Tag icon to open a modal for labeling. The annotation system supports multiple nested labels within a primary label.
Scanning Images
The scanning feature enables automated product image capture with dynamic label assignment. Configure settings for delay, image count, and brightness levels. Position the product, draw labels, capture images, and upload them to the AI model.
Top Scan Tips
Best practices for accurate annotations
Use good lighting to ensure high-quality images, as poor lighting can reduce the accuracy of AI detection.
Label images accurately, positioning labels near the product's edge to clearly define the product boundaries for AI recognition.
Capture images from all angles and sides of the product to improve detection speed and accuracy across various perspectives.
For size or distance detection, use the frame-based detection to position the product within a designated area, ensuring precise measurements.
Avoid excessive use of extra labels, as they increase processing demands. Use a single label for distinct products to optimize resource usage.
Regularly review and update annotations to maintain consistency with product changes, ensuring the AI model remains accurate over time.
Access Control
Role-based permissions for annotation management
| Role | View Annotations | Edit Annotations | Manage Annotations |
|---|---|---|---|
Admin | |||
Branch Manager | |||
Product Operator | |||
Operator | |||
Viewer |
Access to the Annotation Products section is restricted to Admins and Product Operators, ensuring precise and secure maintenance of the AI-driven detection system.