> ## Documentation Index
> Fetch the complete documentation index at: https://help.dingtalk.io/llms.txt
> Use this file to discover all available pages before exploring further.

# AI Image Upload

> Learn how the Image Upload component automatically recognizes text, graphics, and tables in images, and combines advanced AI for scene understanding and logical judgment. Key benefits include proactive risk alerts, end-to-end automation, and high-precision scene adaptation. Common use cases include recognizing price tags, receipts, and product appearance information, as well as extracting passport data.

## Overview

This component not only recognizes text, graphics, tables, and ID information in images, but also integrates advanced AI technologies such as **image classification, object detection, and semantic segmentation** to perform scene understanding and logical judgment on image content. For example, it can identify complex scenarios such as fire smoke, helmet compliance, and abnormal equipment states. It extracts key information from uploaded images and automatically fills the corresponding form fields, elevating the process from "information extraction" to "intelligent decision-making." Key advantages of this feature include:

* Proactive risk alerts

* End-to-end automation

* High-precision scene adaptation

<Note>
  **Basic capabilities of the Image Upload component:** [**View details**](/yida/form/id1g3k)
</Note>

## **Use Cases**

### AI Image Information Extraction

This feature extracts key information from uploaded images and outputs it as fields. Typical scenarios include:

* Price tags: Extract retail price, member price, product name, and specifications from price tag images uploaded by stores.
* Receipts: Extract product name, quantity, specification, unit price, and amount from receipt images.
* Product appearance information: Recognize batch numbers, production dates, and QR Code information on product packaging.
* Passports: Extract name, passport number, and validity period from passport images.

### AI Image Recognition and Analysis

This feature identifies specific elements in images to verify that they meet design requirements or layout standards. Typical scenarios include:

* Packaging design verification: Check whether packaging designs contain errors, such as mirrored logos or missing website addresses.
* Merchandising layout inspection: Evaluate whether store layouts meet standards, including promotion tables, food tasting areas, and product display areas.
* Chain store campaign inspection: Inspect campaign setups in stores, such as banners, LED screens, and posters.

## **Procedure**

### Entry Point

<Steps>
  <Step title="Path: Go to the form page, click the Image Upload AI component >> Attributes panel on the right >> AI intelligent recognition and fill" />

  <Step title="Turn on AI, and then click Image intelligent recognition." />
</Steps>

### Write the Prompt

Keep the following in mind when writing a prompt:

<Steps>
  <Step title="Step 1">
    Describe the business scenario and the information to extract from the image.
  </Step>

  <Step title="Step 2">
    The system provides more than 200 built-in prompt templates covering 15 industries. Click to quickly access scenario templates.
  </Step>
</Steps>

3. We recommend that the prompt include the following:

### Optimize the Prompt

1. Auto-tuning: Click **Optimize and generate form fields**. AI refines the prompt to make it more professional and standardized.

2. Multi-round continuous tuning:

3. To add, modify, or delete fields, append the following instructions at the end of the prompt:

<Warning>
  * ‼️ Tuning examples, in natural language, for your reference: - Add a field: Add a "Production date" field with the data type "Date type." - Rename a field: Rename the "Product name" field to "Item name." - Delete a field: Remove the "Product name" field. - Adjust options of a status field: Change the options of the "Status" field to "Shipped," "Received," and "Cancelled." - Change the field type to a subform: The image contains multiple types of clothing rather than a single item, so change the output field to a "Clothing list" that includes clothing type, size, color, price, and other information.
</Warning>

### Configure the Form Field List

<table>
  <tr>
    <td>**Illustration**</td>
    <td>**Configuration item**</td>
    <td>**Description**</td>
  </tr>

  <tr>
    <td rowSpan={3} />

    <td>**Field name**</td>
    <td>The field name recognized from the image by AI.<br />Optimizing the prompt automatically generates a form field list. Based on the prompt, AI identifies the fields to extract from the image, infers the data type of each field, and recommends a default component type accordingly.</td>
  </tr>

  <tr>
    <td>**Form field**</td>
    <td>Select the form field to fill with the recognition result.<br />Each field automatically matches an existing form field with the same data type and name. If a match exists, the field is bound directly; otherwise, a new form field is created.</td>
  </tr>

  <tr>
    <td>**Component type**</td>
    <td>The component type corresponding to the form field. This can only be selected when adding a new field.<br />For newly added fields, select an appropriate component type as needed. If the new field uses a component type such as subform, Select, or Multiselect, the corresponding subform columns, options, and multiple-choice settings are automatically generated. After saving, the new field is added to the form page automatically.</td>
  </tr>
</table>

### Test the Recognition Results

1. Click to upload an image. The system automatically runs recognition and displays the results.

2. Click **Re-recognize** to run recognition on the image again.

3. Click **Confirm** to save the configuration. Click **Preview** to preview the AI Image Upload component.

## Supported Component Types

The following table shows whether each component supports batch data export:

<table>
  <tr>
    <td>**Component type**</td>
    <td>**Component name**</td>
    <td>**Export supported**</td>
    <td>**Data type description**</td>
  </tr>

  <tr>
    <td rowSpan={17}>**Common components**</td>
    <td>Single-line text</td>
    <td><ul><li>Supported</li></ul></td>
    <td>Text format</td>
  </tr>

  <tr>
    <td>Multi-line text</td>
    <td><ul><li>Supported</li></ul></td>
    <td>Text format</td>
  </tr>

  <tr>
    <td>Numerical</td>
    <td><ul><li>Supported</li></ul></td>
    <td>Number</td>
  </tr>

  <tr>
    <td>Rating</td>
    <td><ul><li>Supported</li></ul></td>
    <td>Number</td>
  </tr>

  <tr>
    <td>Select</td>
    <td><ul><li>Supported</li></ul></td>
    <td>Text format</td>
  </tr>

  <tr>
    <td>Checkbox</td>
    <td><ul><li>Supported</li></ul></td>
    <td>Text format; multiple options are separated by commas</td>
  </tr>

  <tr>
    <td>Date range</td>
    <td><ul><li>Supported</li></ul></td>
    <td>Date format</td>
  </tr>

  <tr>
    <td>Date</td>
    <td><ul><li>Supported</li></ul></td>
    <td>Date format</td>
  </tr>

  <tr>
    <td>User</td>
    <td><ul><li>Supported</li></ul></td>
    <td>Text format</td>
  </tr>

  <tr>
    <td>Department</td>
    <td><ul><li>Supported</li></ul></td>
    <td>Text format</td>
  </tr>

  <tr>
    <td>Attachment</td>
    <td>❌ Not supported</td>
    <td>/</td>
  </tr>

  <tr>
    <td>Word template</td>
    <td>❌ Not supported</td>
    <td>/</td>
  </tr>

  <tr>
    <td>Signature</td>
    <td>❌ Not supported</td>
    <td>/</td>
  </tr>

  <tr>
    <td>Smart invoice</td>
    <td>❌ Not supported</td>
    <td>/</td>
  </tr>

  <tr>
    <td>ID number</td>
    <td>❌ Not supported</td>
    <td>/</td>
  </tr>

  <tr>
    <td>Phone number</td>
    <td>❌ Not supported</td>
    <td>/</td>
  </tr>

  <tr>
    <td>Subform</td>
    <td><ul><li>Supported</li></ul></td>
    <td>Text format</td>
  </tr>

  <tr>
    <td rowSpan={15}>**Premium components**</td>
    <td>Dropdown Select</td>
    <td><ul><li>Supported</li></ul></td>
    <td>Text format</td>
  </tr>

  <tr>
    <td>Dropdown Checkbox</td>
    <td><ul><li>Supported</li></ul></td>
    <td>Text format; multiple options are separated by commas</td>
  </tr>

  <tr>
    <td>Cascade Select</td>
    <td><ul><li>Supported</li></ul></td>
    <td>Text format</td>
  </tr>

  <tr>
    <td>Country/Region</td>
    <td><ul><li>Supported</li></ul></td>
    <td>Text format</td>
  </tr>

  <tr>
    <td>Address</td>
    <td><ul><li>Supported</li></ul></td>
    <td>Text format</td>
  </tr>

  <tr>
    <td>Location</td>
    <td>❌ Not supported</td>
    <td>/</td>
  </tr>

  <tr>
    <td>Associated form</td>
    <td><ul><li>Supported</li></ul></td>
    <td>Text format</td>
  </tr>

  <tr>
    <td>Serial number</td>
    <td>❌ Not supported</td>
    <td>/</td>
  </tr>

  <tr>
    <td>Organization</td>
    <td>❌ Not supported</td>
    <td>/</td>
  </tr>

  <tr>
    <td>UnionPay</td>
    <td>❌ Not supported</td>
    <td>/</td>
  </tr>

  <tr>
    <td>Budget control</td>
    <td>❌ Not supported</td>
    <td>/</td>
  </tr>

  <tr>
    <td>Group</td>
    <td>❌ Not supported</td>
    <td>/</td>
  </tr>

  <tr>
    <td>Associated query</td>
    <td>❌ Not supported</td>
    <td>/</td>
  </tr>

  <tr>
    <td>Rich Text</td>
    <td>❌ Not supported</td>
    <td>/</td>
  </tr>

  <tr>
    <td>Image and text display</td>
    <td>❌ Not supported</td>
    <td>/</td>
  </tr>
</table>

<Note>
  **Notes**

  1. The maximum image size is 20 MB.
  2. Supported image formats: PNG and JPG.
  3. The more detailed and accurate the prompt, the better the recognition results.
  4. When you add a new field, the component type is recommended automatically based on the data format.
  5. Recognition results appear in real time in the right panel.
</Note>

## FAQ

**Q: Why are the recognition results inaccurate?**
A: Check whether the prompt is clear enough. Use the tuning feature to refine it if needed.

**Q: How do I choose the right component type?**
A: The system recommends a suitable component type based on the recognized data format. Select one from the recommended list.

**Q: Nothing happens after I upload an image.**
A: Make sure the image format is correct and the size is within the limit. Also check that the network connection is working properly.