Dev Guide
Model Parameter Specification
Model Parameter is a critical definition in Cinfer’s model management. It governs the external input/output schema for ONNX models and serves as the core configuration file defining how a model interacts with the outside world. It describes model configuration, required input data formats, and the structure of inference outputs. The file uses YAML and follows a JSON-Schema-like convention with three top-level fields: config, inputs, and outputs. Details are below.
config (object, optional)
The config field defines general model configuration and behavior.
| Key | Type | Description | Example |
|---|---|---|---|
processing_strategy | string | Defines preprocessing and postprocessing strategy. generic uses a common pipeline. | processing_strategy: generic |
labels | map | Label mapping | labels:<br> 0: "person"<br> 1: 'bicycle'<br> 2: 'car' |
type | string | Model type: object detection, pose estimation, instance segmentation, semantic segmentation | type: "object detection" |
shapes | array | Model input shape | shapes: [1, 3, 640, 640] |
post_format | array | Built-in postprocessing output helpers | concatLabels |
... | ... | Future extensible configs as needed. |
inputs (array, required)
An array defining one or more input parameters required for inference. Each element is an object describing a single input parameter.
outputs (array, required)
An array defining the output data structure after inference. Each element is an object describing a single output parameter.
inputs and outputs element definition
Each element of inputs and outputs is a parameter definition object describing a single input or output parameter.
Common properties
| Property | Type | Required? | Description |
|---|---|---|---|
| name | string | yes | Unique parameter name used in code. |
| description | string | yes | Detailed description of the parameter. |
| type | string | yes | Data type. Supports string, float, integer, boolean, array, object. |
| required | boolean | no | Whether this parameter is required. Defaults to false if omitted. |
string type
| Property | Type | Description | Example |
|---|---|---|---|
format | string | Specific string formats. Use | to separate multiple acceptable formats. Common values: uri (URL), base64 (Base64-encoded string). | uri | base64 |
minLength | integer | Minimum string length. | minLength: 1 |
maxLength | integer | Maximum string length. | maxLength: 1024 |
pattern | string | Regex pattern the string must match. | pattern: "^[a-zA-Z0-9_-]+$" |
float and integer types
| Property | Type | Description | Example |
|---|---|---|---|
default | float or integer | Default value if required is false and the parameter is missing. | default: 0.25 |
minimum | float or integer | Minimum allowed value (inclusive). | minimum: 0.0 |
maximum | float or integer | Maximum allowed value (inclusive). | maximum: 1.0 |
array type
| Property | Type | Description | Example |
|---|---|---|---|
items | object | Required. A parameter definition object describing each array element. | See nested example below |
minItems | integer | Minimum number of elements. | minItems: 1 |
maxItems | integer | Maximum number of elements. | maxItems: 100 |
object type
| Property | Type | Description | Example |
|---|---|---|---|
properties | object | Required. Keys are property names and values are parameter definition objects for those properties. | See nested example below |
Definition examples
yolov8-seg (instance segmentation)
config:
processing_strategy: generic
labels:
0: 'person'
1: 'bicycle'
2: 'car'
3: 'motorcycle'
4: 'airplane'
5: 'bus'
6: 'train'
7: 'truck'
8: 'boat'
9: 'traffic light'
10: 'fire hydrant'
11: 'stop sign'
12: 'parking meter'
13: 'bench'
14: 'bird'
15: 'cat'
16: 'dog'
17: 'horse'
18: 'sheep'
19: 'cow'
20: 'elephant'
21: 'bear'
22: 'zebra'
23: 'giraffe'
24: 'backpack'
25: 'umbrella'
26: 'handbag'
27: 'tie'
28: 'suitcase'
29: 'frisbee'
30: 'skis'
31: 'snowboard'
32: 'sports ball'
33: 'kite'
34: 'baseball bat'
35: 'baseball glove'
36: 'skateboard'
37: 'surfboard'
38: 'tennis racket'
39: 'bottle'
40: 'wine glass'
41: 'cup'
42: 'fork'
43: 'knife'
44: 'spoon'
45: 'bowl'
46: 'banana'
47: 'apple'
48: 'sandwich'
49: 'orange'
50: 'broccoli'
51: 'carrot'
52: 'hot dog'
53: 'pizza'
54: 'donut'
55: 'cake'
56: 'chair'
57: 'couch'
58: 'potted plant'
59: 'bed'
60: 'dining table'
61: 'toilet'
62: 'tv'
63: 'laptop'
64: 'mouse'
65: 'remote'
66: 'keyboard'
67: 'cell phone'
68: 'microwave'
69: 'oven'
70: 'toaster'
71: 'sink'
72: 'refrigerator'
73: 'book'
74: 'clock'
75: 'vase'
76: 'scissors'
77: 'teddy bear'
78: 'hair drier'
79: 'toothbrush'
type: "instance segmentation"
shapes: [1, 3, 640, 640]
inputs:
- name: "image"
description: "The image can be a valid URL or a Base64 string in data URI format"
type: "string"
format: "uri | base64"
required: true
- name: "conf_threshold"
description: "Optional. Confidence threshold (keep detections with confidence ≥ this)"
type: "float"
required: false
default: 0.25
minimum: 0.0
maximum: 1.0
- name: "nms_threshold"
description: "Optional. NMS threshold (suppress boxes with IoU > this value)"
type: "float"
required: false
default: 0.45
minimum: 0.0
maximum: 1.0
outputs:
- name: "data"
description: "Detection results for one or multiple images; each element corresponds to one image and its objects"
type: "array"
required: true
items:
description: "Describes a single image and its detection results"
type: "object"
required: true
properties:
file_name:
description: "Recognized file index or name"
type: "string"
required: true
detections:
description: "All detected objects in the image"
type: "array"
items:
type: "object"
properties:
boxes:
type: "array"
description: "Bounding box [x, y, w, h]"
items:
type: "integer"
minLength: 4
maxLength: 4
conf:
description: "Confidence score (0.0 to 1.0)"
type: "float"
required: true
cls:
description: "Predicted class label (e.g., '0', 'person', 'dog')"
type: "string"
required: true
masks:
type: "array"
description: "Optional: segmentation polygon points; each sub-array is a vertex [x, y]"
items:
type: "array"
minItems: 2
maxItems: 2
items:
type: "integer"
description: "Pixel coordinate, non-negative"
minimum: 0
points:
type: "array"
description: "Optional: pose keypoints; each is [x, y, id, conf]"
items:
type: "array"
description: "x/y are pixel coordinates (float), id is keypoint index (int), conf is confidence (float)"
minItems: 4
maxItems: 4
skeleton:
type: "array"
description: "Skeleton edges; each item is a pair of keypoint indices [idx1, idx2]; returned with points"
items:
type: "array"
description: "Each edge consists of two keypoint indices"
items:
type: "integer"
description: "Keypoint index"
minItems: 2
maxItems: 2
Pointer gauge reading example (instance segmentation)
config:
processing_strategy: pointer
labels:
0: "0"
1: "1"
type: "instance segmentation"
shapes: [1, 3, 640, 640]
inputs:
- name: "image"
description: "The image can be a valid URL or a Base64 string in data URI format"
type: "string"
format: "uri | base64"
required: true
- name: "conf_threshold"
description: "Optional. Confidence threshold (keep detections with confidence ≥ this)"
type: "float"
required: false
default: 0.25
minimum: 0.0
maximum: 1.0
- name: "nms_threshold"
description: "Optional. NMS threshold (suppress boxes with IoU > this value)"
type: "float"
required: false
default: 0.45
minimum: 0.0
maximum: 1.0
- name: "scale_min"
description: "Scale start value"
type: "float"
required: true
default: 0.0
minLength: 4
maxLength: 4
- name: "scale_max"
description: "Scale end value"
type: "float"
required: true
default: 60.0
minLength: 4
maxLength: 4
outputs:
- name: "data"
description: "Detection results for one or multiple images; each element corresponds to one image and its objects"
type: "array"
required: true
items:
description: "Describes a single image and its detection results"
type: "object"
required: true
properties:
file_name:
description: "Recognized file index or name"
type: "string"
required: true
detections:
description: "All detected objects in the image"
type: "array"
items:
type: "object"
properties:
boxes:
type: "array"
description: "Bounding box [x, y, w, h]"
items:
type: "integer"
minLength: 4
maxLength: 4
conf:
description: "Confidence score (0.0 to 1.0)"
type: "float"
required: true
cls:
description: "Predicted class label (e.g., '0', 'person', 'dog')"
type: "string"
required: true
masks:
type: "array"
description: "Optional: segmentation polygon points; each sub-array is a vertex [x, y]"
items:
type: "array"
minItems: 2
maxItems: 2
items:
type: "integer"
description: "Pixel coordinate, non-negative"
minimum: 0
points:
type: "array"
description: "Optional: pose keypoints; each is [x, y, id, conf]"
items:
type: "array"
description: "x/y are pixel coordinates (float), id is keypoint index (int), conf is confidence (float)"
minItems: 4
maxItems: 4
skeleton:
type: "array"
description: "Skeleton edges; each item is a pair of keypoint indices [idx1, idx2]; returned with points"
items:
type: "array"
description: "Each edge consists of two keypoint indices"
items:
type: "integer"
description: "Keypoint index"
minItems: 2
maxItems: 2
Water meter reading example (object detection)
config:
processing_strategy: generic
labels:
0: "0"
1: "1"
2: "2"
3: "3"
4: "4"
5: "5"
6: "6"
7: "7"
8: "8"
9: "9"
type: "object detection"
shapes: [1, 3, 640, 640]
post_format:
- "concatLabels"
inputs:
- name: "image"
description: "The image can be a valid URL or a Base64 string in data URI format"
type: "string"
format: "uri | base64"
required: true
- name: "conf_threshold"
description: "Optional. Confidence threshold (keep detections with confidence ≥ this)"
type: "float"
required: false
default: 0.25
minimum: 0.0
maximum: 1.0
- name: "nms_threshold"
description: "Optional. NMS threshold (suppress boxes with IoU > this value)"
type: "float"
required: false
default: 0.45
minimum: 0.0
maximum: 1.0
outputs:
- name: "data"
description: "Detection results for one or multiple images; each element corresponds to one image and its objects"
type: "array"
required: true
items:
description: "Describes a single image and its detection results"
type: "object"
required: true
properties:
file_name:
description: "Recognized file index or name"
type: "string"
required: true
detections:
description: "All detected objects in the image"
type: "array"
items:
type: "object"
properties:
boxes:
type: "array"
description: "Bounding box [x, y, w, h]"
items:
type: "integer"
minLength: 4
maxLength: 4
conf:
description: "Confidence score (0.0 to 1.0)"
type: "float"
required: true
cls:
description: "Predicted class label (e.g., '0', 'person', 'dog')"
type: "string"
required: true
masks:
type: "array"
description: "Optional: segmentation polygon points; each sub-array is a vertex [x, y]"
items:
type: "array"
minItems: 2
maxItems: 2
items:
type: "integer"
description: "Pixel coordinate, non-negative"
minimum: 0
points:
type: "array"
description: "Optional: pose keypoints; each is [x, y, id, conf]"
items:
type: "array"
description: "x/y are pixel coordinates (float), id is keypoint index (int), conf is confidence (float)"
minItems: 4
maxItems: 4
skeleton:
type: "array"
description: "Skeleton edges; each item is a pair of keypoint indices [idx1, idx2]; returned with points"
items:
type: "array"
description: "Each edge consists of two keypoint indices"
items:
type: "integer"
description: "Keypoint index"
minItems: 2
maxItems: 2
Built-in OCR model definition example
config:
inputs:
- name: "image"
description: "The image can be a valid URL or a Base64 string in data URI format"
type: "string"
format: "uri | base64"
required: true
- name: "conf_threshold"
description: "Optional. Confidence threshold (keep detections with confidence ≥ this)"
type: "float"
required: false
default: 0.5
minimum: 0.0
maximum: 1.0
outputs:
- name: "data"
description: "Detection results for one or multiple images; each element corresponds to one image and its detections"
type: "array"
required: true
items:
description: "Describes a single image and its detection results"
type: "object"
required: true
properties:
file_name:
description: "Recognized file index or name"
type: "string"
required: true
detections:
description: "All detected objects in the image"
type: "array"
items:
type: "object"
properties:
box:
type: "array"
description: "Bounding box [x, y, w, h]"
items:
type: "integer"
minLength: 4
maxLength: 4
conf:
description: "Confidence score (0.0 to 1.0)"
type: "float"
required: true
cls:
description: "Recognized text, e.g., '(Thirty-Six Stratagems Full Interpretation)'"
type: "string"
required: true