Home

Defect Detection API Documentation

1. Connection Test

GET /api/connection_test

Verify API service availability.

Example Request

curl https://api.defect-ai.com/api/connection_test

Success Response

{
    "status": "ok",
    "message": "connection successful"
}

2. Defect Detection Interface

POST /api/interface

Submit product images for defect detection and analysis.

Request Parameters

Parameter Type Description
product_code Integer Product type identifier (required in form data)
images File[] Array of image files (supported formats: JPEG, PNG, etc.)

Example Request

curl -X POST \
  -F "product_code=1" \
  -F "images=@20250222_103428260_0000000001_T0000_01_CAM.png" \
  -F "images=@20250222_103428260_0000000001_T0000_02_CAM.png" \
  -F "images=@20250222_103428260_0000000001_T0000_03_CAM.png" \
  -F "images=@20250222_103428260_0000000001_T0000_04_CAM.png" \
  -F "images=@20250222_103428260_0000000001_T0000_05_CAM.png" \
  -F "images=@20250222_103428260_0000000001_T0000_06_CAM.png" \
  https://api.defect-ai.com/api/interface

Success Response

{
  "message": "Inference completed.",
  "result": {
    "defect_percent": 66.667,
    "defect_type": "1-1",
    "defect_type_name": "crack",
    "interface_result": ["1-1", "1-1", "1-0", "1-1", "1-0", "1-1"],
    "interface_result_name": ["crack", "crack", "No defect", "crack", "No defect", "crack"],
    "interface_time": 1.07,
    "is_defect": true,
    "product_name": "syringe",
    "product_type": 1,
    "threshold_percent": 20.0
  },
  "status": "success"
}

Error Responses

Code Response
400 
{"error": "Missing images in request"}
{"error": "Invalid product_code"}
500 
{"error": "Inference failed: [error description]"}

Response Field Definitions

Field Type Description
is_defect Boolean Overall defect status (true if defect percentage ≥ threshold)
defect_type String Dominant defect type in format "product_type-defect_code"
defect_type_name String Human-readable name of the dominant defect type (e.g., "crack")
interface_result Array Raw defect codes detected in all images
interface_result_name Array Human-readable defect names corresponding to each code in interface_result
interface_time Float Time taken for inference processing in seconds
defect_percent Float Percentage of defective items relative to valid items
product_name String Human-readable name of the product being analyzed
product_type Integer Product type identifier provided in the request
threshold_percent Float Current threshold percentage used for determination

Notes