Image Re-photography Detection

Tool Introduction

"Image Re-photography Detection" is an intelligent online tool designed to help users quickly determine whether an uploaded image is a re-photographed image. The tool uses advanced image analysis algorithms to identify potential re-photography traces in images, such as screen textures, light reflections, distortions, etc., and outputs a confidence score. To optimize the detection effect, the tool provides flexible image pre-processing and scaling functions. Users can choose "size scaling" or "ratio scaling" to pre-process the image as needed to meet the requirements of the detection model, thereby improving the accuracy of re-photography detection.

How to Use

1. Upload Image: Click the "Upload Image" area and select an image you need to identify (supports .png, .jpg formats, size not exceeding 5MB).
2. Select Scaling Type (Optional): According to your needs, select "Size Scaling" or "Ratio Scaling" in "Scaling Type" as a pre-processing operation before identification.
   • If "Size Scaling" is selected: You can enter the target "Width (px)" and "Height (px)", and choose whether to "Keep Aspect Ratio" to adjust the image size.
   • If "Ratio Scaling" is selected: You can enter a "Scaling Ratio" (0-2) to proportionally enlarge or reduce the image.
3. Submit for Identification: Click the "Identify" button, and the tool will process the image and perform re-photography detection.
4. View Results: After identification is complete, the result area will display the identified image type (e.g., "Re-photographed Image", "Original Image") and its corresponding "Confidence".

Usage Example

Suppose you have an image and suspect it was re-photographed from a screen, and you want to verify it with this tool.

• Example Input Data:
• Upload Image: An image file named screenshot_photo.jpg (assuming 4MB in size).
• Scaling Type: Select "Size Scaling".
• Width (px): Enter 800.
• Height (px): Enter 600.
• Keep Aspect Ratio: Select "Yes".
• Specific Operation Demonstration:
  1. In the "Upload Image" area, click and select screenshot_photo.jpg.
  2. In the "Scaling Type" dropdown menu, select "Size Scaling".
  3. In the "Width (px)" input box, fill in 800.
  4. In the "Height (px)" input box, fill in 600.
  5. "Keep Aspect Ratio" is selected "Yes" by default, keep it unchanged.
  6. Click the "Identify" button at the bottom.
• Expected Output Result:

  After the tool processes, the result area may display:

  Category Name	Confidence
  Re-photographed Image	0.92

  This means the tool determines the image to be a re-photographed image with 92% confidence.

Frequently Asked Questions

• Q: What is image re-photography detection? A: Image re-photography detection is an image analysis technology used to detect whether a digital image was created by physically photographing another digital display content (such as a computer screen, TV, printed photo, etc.) with a physical device (such as a mobile phone, camera). This helps verify the authenticity of the image source.
• Q: What image formats and size limits are supported? A: Currently, PNG and JPG format images are supported for upload, with a single image size not exceeding 5MB.
• Q: What is the purpose of the pre-processing scaling function? A: The pre-processing scaling function allows you to adjust the image size before identification. This helps standardize the specifications of input images and can sometimes improve the performance and accuracy of the identification model, especially when dealing with images of different resolutions or sizes.
• Q: What does the output confidence represent? A: Confidence is a value between 0 and 1, indicating the model's certainty about the identification result. For example, a confidence of 0.92 means the model is 92% confident that the image belongs to the "re-photographed image" or "original image" category. The higher the confidence, the more reliable the result.

Notes

• Please ensure that the uploaded image is in PNG or JPG format and its size does not exceed 5MB.
• The setting of pre-processing scaling parameters will affect the identification results. It is recommended to adjust them based on image characteristics and experience. If unsure, you can use the default settings or no scaling.
• The confidence of the identification result is for reference only and does not guarantee 100% absolute accuracy. In critical application scenarios, it is recommended to combine with other information for comprehensive judgment.
• This tool is mainly optimized for common screen re-photography scenarios, and its effect may be limited for other complex image forgery situations.