Object Detection & Classification

Show info about photo: objects, faces, give the title to the picture and detect pose if applicable.

Request parameters

Description
Parameter’s name Parameter’s type Description
raw_input boolean False - Use Tensorflow protobuf record format fot feeding data to model, True - Use numpy arrays.
out_mime_type string Optional. Used on a client side to identify content nature.
out_filters string Optional. Comma separated list of model outputs to return.
input bytes image to analyze
detect_objects bool Detect objects and tag image
detect_faces bool Detect faces on image
detect_poses bool Detect poses for persons on image
build_caption bool Build caption for image
form-data
curl -X POST \
    -H 'Authorization: Bearer YOUR_USER_TOKEN' \
    -H 'Content-Type: multipart/form-data' \
    -F "out_filters=output" \
    -F "raw_input=true" \
    -F "out_mime_type=image/png" \
    -F "bytes_input=@PATH_TO_INPUT" \
    -F "bool_detect_objects=detect_objects_value" \
    -F "bool_detect_faces=detect_faces_value" \
    -F "bool_detect_poses=detect_poses_value" \
    -F "bool_build_caption=build_caption_value" \
    https://cloud.kibernetika.io/api/v0.2/workspace/hpc/serving/photo-inside-picture/tfproxy/9000/any
json
curl -X POST \
    -H 'Authorization: Bearer YOUR_USER_TOKEN' \
    -H 'Content-Type: application/json' \
    -d '
      {
        "inputs": {
          "build_caption": {
            "data": true,
            "dtype": 10
          },
          "detect_faces": {
            "data": true,
            "dtype": 10
          },
          "detect_objects": {
            "data": true,
            "dtype": 10
          },
          "detect_poses": {
            "data": true,
            "dtype": 10
          },
          "input": {
            "data": "base64_encoded_input_contents",
            "dtype": 7
          }
        },
        "out_filters": "output",
        "out_mime_type": "image/png",
        "raw_input": true
      }
    '
    https://cloud.kibernetika.io/api/v0.2/workspace/hpc/serving/photo-inside-picture/tfproxy/9000/any

Response format

{
  "output": [
    "bytes"
  ],
  "table_output": [
    "string"
  ]
}