跳到主要内容

Image Quantifier Tutorial

Overview​

The Image Quantifier runs in a Hugging Face Space (Gradio) and is embedded into the web app. It performs server‑side analysis on uploaded images of leaves or seeds/grains, computes morphology metrics, and returns an overlay preview plus a CSV. Camera capture depends on the Space implementation and is not guaranteed in the embedded mode; use local files when the camera option is unavailable.

Key Features​

  • Leaf/Seed Quantification: Automated detection and per‑sample metrics
  • Reference Scale: Coin/ruler/no‑reference modes; ruler requires ref_size_mm
  • Expected Count: Optionally limit analyzed components to a target count
  • Color Segmentation: HSV range (H low/high) with color tolerance; area filters
  • Overlay Preview: Server‑generated image with bounding boxes and markers
  • CSV Export: Per‑component measurements for downstream analysis

Quick Start​

1. Access the Application​

Visit in your browser: /app/image

2. System Requirements​

  • Modern Web Browser: Chrome, Firefox, Safari, or Edge with camera support
  • Image Files: JPEG or PNG format with sufficient resolution
  • Camera Access: For live image capture functionality
  • Adequate Lighting: Consistent illumination for accurate measurements

Detailed Usage Steps​

Step 1: Sample Information Setup​

  1. Sample ID Assignment

    • Enter unique identifier for each sample set
    • Use descriptive names for easy reference
    • Maintain consistent naming conventions

  2. Sample Count Specification

    • Estimate number of samples in the image
    • System uses this for initial processing optimization
    • Can be adjusted during analysis if needed

Step 2: Image Acquisition​

  1. Image Upload

    • Click the upload control in the embedded Space to select files
    • Supported formats: JPEG, PNG (depending on Space)
    • Use adequate resolution and consistent lighting

  2. Camera Capture (Optional)

    • If the Space has a camera widget, you can capture a photo
    • In embedded mode this option may be disabled; prefer file upload

Step 3: Analysis Configuration​

  1. Sample Type

    • Choose "leaves" or "seeds/grains" for tailored sorting

  2. Reference Mode & Size

    • Select none / coin / ruler; set ref_size_mm when using a ruler

  3. Segmentation & Filters

    • Set HSV H‑range (low_h, high_h) and color_tol
    • Set min_area_px/max_area_px to filter small/large components
    • Optionally set expected_count

Step 4: Processing & Preview​

  1. Automatic Detection

    • The Space segments components and computes per‑component metrics

  2. Overlay Preview

    • Review the generated overlay image to verify segmentation

Step 5: Results Export​

  1. Measurement Data

    • Download CSV with component metrics (area, perimeter, axes, etc.)

  2. Overlay Image

    • Save the annotated overlay image for documentation

Technical Specifications​

Image Requirements​

  • Format: JPEG, PNG
  • Resolution: Minimum 640×480 pixels, recommended 1920×1080 or higher
  • Color Depth: 8-bit or higher for accurate color analysis
  • Compression: Minimal compression for measurement accuracy

Measurement Parameters​

Geometric Measurements​

  • Area: Square millimeters or square centimeters
  • Perimeter: Millimeters with sub-pixel accuracy
  • Major/Minor Axis: Length of longest and shortest dimensions
  • Aspect Ratio: Ratio of major to minor axis

Shape Descriptors​

  • Circularity: 4π × Area / Perimeter²
  • Solidity: Area / Convex Hull Area
  • Form Factor: Various shape complexity measures
  • Roundness: Compactness relative to circle

Color Analysis​

  • RGB Channels: Individual color channel intensities
  • HSV Values: Hue, saturation, and value components
  • NDVI Estimation: Normalized Difference Vegetation Index
  • Chlorophyll Index: Relative chlorophyll content estimation

Accuracy Specifications​

  • Spatial Resolution: Dependent on image resolution and reference scale
  • Measurement Precision: Typically ±1-2 pixels
  • Repeatability: Coefficient of variation < 5% for standard conditions
  • Calibration: Requires proper reference scale placement

Best Practices​

Image Acquisition​

  1. Lighting Conditions

    • Use consistent, diffuse lighting
    • Avoid shadows and specular reflections
    • Maintain uniform illumination across samples

  2. Camera Settings

    • Use manual focus for consistent sharpness
    • Set appropriate white balance
    • Avoid digital zoom for measurement accuracy
    • Use tripod for stability

  3. Sample Preparation

    • Ensure samples are flat and properly oriented
    • Avoid overlapping or touching samples
    • Use neutral background for contrast
    • Keep samples clean and dry

Measurement Validation​

  1. Reference Scale

    • Use standardized reference objects
    • Place reference in same plane as samples
    • Verify reference dimensions are accurate
    • Include reference in every image

  2. Quality Control

    • Check for consistent measurement units
    • Verify sample count matches expectations
    • Review boundary detection accuracy
    • Validate against manual measurements

Data Management​

  1. File Organization

    • Use descriptive file naming conventions
    • Maintain metadata with each analysis
    • Archive original images with results
    • Version control for analysis parameters

  2. Statistical Analysis

    • Use appropriate statistical methods
    • Account for measurement uncertainty
    • Consider sample size requirements
    • Document analysis methodology

Troubleshooting​

Common Issues​

1. Poor Sample Detection

  • Check image contrast and lighting
  • Verify sample-background differentiation
  • Adjust detection sensitivity if available
  • Consider manual sample boundary adjustment

2. Inaccurate Measurements

  • Verify reference scale placement and accuracy
  • Check image resolution and focus quality
  • Ensure samples are in same plane as reference
  • Review camera calibration if available

3. Camera Access Problems

  • Grant camera permissions in browser
  • Check if other applications are using camera
  • Verify camera hardware functionality
  • Try different browser if issues persist

Performance Optimization​

For Large Images

  • Use appropriate image resolution for required accuracy
  • Consider image compression for faster processing
  • Process images in batches if multiple analyses needed

For Complex Samples

  • Use higher resolution images for detailed features
  • Consider multiple imaging angles if 3D information needed
  • Use specialized lighting for challenging samples

Technical Support​

If you encounter technical issues:

  1. Check browser console for error messages
  2. Verify image format and size requirements
  3. Ensure camera permissions are granted
  4. Contact support with specific error details and sample images

Author: Liangchao Deng, Ph.D. Candidate, Shihezi University / CAS-CEMPS
This tutorial applies to Image Quantifier v1.0 Optimized for plant biology and agricultural research applications