Importing local image processing

If your original data is images, follow this document for processing.

Select 'From picture' and 'New Task' in the 'Import Data Source' section on the dataset details page.

All imports are referred to as tasks, where each task can include multiple similar data entries for processing. This section on local image tasks allows for the addition of multiple image data entries.

1. Create a New Task

Create a new task to import images from local for the dataset

On the task creation page, provide a name for your task (up to 20 characters). This name will assist you in quickly locating and managing this task in the task list.

2. Upload Local Image Files

1. Click on the upload area, drag and drop your local files into the upload box, or click the upload button to select files for uploading.

2. Supported file formats include: .jpg、.png.

3. You can upload up to 50 files per task, with each file not exceeding 200MB in size.

4. Ensure that multiple files uploaded within a task have similar content for parameter extraction and output processing.

3. Task Settings

1. Task settings are similar to importing tasks from the web and involve configuring fields and content extraction.

2. Select an appropriate parsing method based on the file type to ensure that the system can correctly process the uploaded files.

4. Fetch Parameters

1. Default Field Types:

   • Get image text: The system will attempt to extract all text information from the image content.

2. Custom Fields:

   • If you need to classify specific extracted data into certain fields, you can click "+ Add Field" and add the field name and description.

   • For example, if there is a nickname in the image that needs to be extracted, the field name key would be: nickname, and the field description: user nickname.

   • Please add in English only. The more detailed the field description, the more accurate the extraction will be.

5. Output Settings

Once you have configured the fetch parameters, you will need to set up output settings to determine how the extracted data will be saved and exported.

1. Output Format Settings

   • You can choose to save the retrieved data in either JSON or Markdown format. JSON format is more suitable for subsequent API program calls, while Markdown format is better suited for knowledge base data processing.

6. Save or Execute Task Immediately

1. Save and manually execute the task later:

   • If you wish to configure the task first without initiating the scraping process immediately, you can click on the "Save and manually execute the task later" button. The task will be saved in the task list for manual initiation at a later time.

2. Execute the task immediately:

   • If you are prepared to instantly scrape webpage data, click the "Execute the task immediately" button. The system will commence data scraping and import it into the specified dataset.