Layout configuration
Layout configuration for Saral App to detect and predict
  1. 1.
    Tools like MS VoTT can be used to tag ROIs with the layout. This will give a raw VoTT Json with ROI coordinates.
2. Use Jupyter Notebook to transform raw VoTT Json to target Layout specification json format.
3. This final layout json should be configured in backend GET /roi/{examId} API for each exam to enable the layout.
Starting from v1.5.0 release each layout can have threshold minimum width,minimum height and detectionRadius configured. This configuration is used by App to detect the sheet only if minimum width and height between dark corner circles is met with scanned sheet/layout. This is to avoid invalid scans which may result in unwanted predictions.
Sample layout with threshould minWidth, minHeight and radius(radius of 4 corner dark circles for alignment) configuration.
1
"layout": {
2
"version": "1.0",
3
"name": "HINDI8S13QOMR Exam Sheet Form",
4
"threshold": {
5
"minWidth" : 500,
6
"minHeight": 200,
7
"detectionRadius": 12
8
},
9
"cells": [
10
{
11
"cellId": "1",
12
"rois": [
13
{
14
"annotationTags": "ROLLNUMBERID1_1",
15
"extractionMethod": "NUMERIC_CLASSIFICATION",
16
"roiId": "1",
17
"index": 0,
18
"rect": {
19
"top": 151,
20
"left": 54,
21
"bottom": 178,
22
"right": 69
23
}
24
},
Copied!
Copy link
Edit on GitHub