Skip to main content

Vision Detect

Given an API call determines the list of items the image contains in descending order of likelihood

How to Get Your Endpoint:​

Go to https://ezout.store/contact and reach out to our team. We should get back to you soon with personal support and a custom endpoint if needed.

Query Params:​

  • upload: 0 (0 for no upload to S3 bucket, 1 for upload to S3 bucket)
  • s3_path: EzOutVision%2FscanImages-dev ()
  • box_conf: 0.9 ()

Body​

form-data:

  • image: /home/my_image.jpg
  • Content-Type: multipart/form-data

API Calls:​

Here is how to call the API in each language:

import requests

url = "https://<example>.com/main?upload=0&s3_path=EzOutVision%2FscanImages-dev&box_conf=0.9"

payload = {}
files=[
  ('image',('tomato-test.png',open('/Users/user/Downloads/image-name.png','rb'),'image/png'))
]
headers = {
  'Content-Type': 'multipart/form-data'
}

response = requests.request("POST", url, headers=headers, data=payload, files=files)

print(response.text)

Output Format:​

{
    {Task: ['inference']}:[
        {detection [x1, y1, x2, y2, probability, class_indice]},
        ...
    ],
    upload_to_s3 (to upload to s3 or not):  int [0,1], 
    s3_path (where it is uploaded to on s3): str,
    box_conf (acceptable confidence of the detection before considered background): float
}

Example Output:​

{
    "inference": [
        [
            37.48759078979492,
            0.0,
            614.691650390625,
            263.84075927734375,
            0.500098466873169,
            16.0
        ]
    ],
    "upload_to_s3": 0,
    "s3_path": "EzOutVision/scanImages-dev",
    "box_conf": 0.9
}

Capabilities​

Items that our models can detect:

0: "beans",
1: "cake",
2: "candy",
3: "cereal",
4: "chips",
5: "chocolate",
6: "coffee",
7: "corn",
8: "fish",
9: "flour",
10: "honey",
11: "jam",
12: "juice",
13: "milk",
14: "nuts",
15: "oil",
16: "pasta",
17: "rice",
18: "soda",
19: "spices",
20: "sugar",
21: "tea",
22: "tomato_sauce",
23: "vinegar",
24: "water"

Model precision by item: