OCR procedure

Official Content
This documentation is valid for:

Extracts text from an image. The acronym OCR means 'Optical Character Recognition'.

Parameters

Configuration

The following table resumes the configuration properties (access credentials) you must set in order to use this AI task.

PropertyKey
ProviderType Id Key SecretKey
Alibaba - 用户AccessKey 用户AccessKey
Amazon - Rekognition Rekognition
Baidu 视觉技术 视觉技术 视觉技术
Google - Cloud Vision API -
IBM - Visual Recognition
(deprecated, dark-beta)		 -
Microsoft - Computer Vision -
MLKit ML Kit API ML Kit API -
SAP - Sandbox Environment
(Deprecated)		 -
Tencent 通用OCR 通用OCR -

Sample

Taking the following image input, the table below shows the text are identified for each provider (as a JSON structure) and the time it takes for processing it.

GeneXusAI - Image module - Sample

Provider Output Benchmark
Alibaba 
[{
	"label": "LevI'S",
	"confidence": 0.999,
	"top": 651,
	"left": 424,
	"width": 128,
	"height": 40
}]
GeneXusAI - Alibaba - OCR
 3156ms
Amazon 
[{
    "label": "LEVI'S",
    "confidence": 0.977,
    "top": 668,
    "left": 428,
    "width": 127,
    "height": 31
}]
GeneXusAI - Amazon - OCR
 13531ms
Baidu 
[{
	"label": "evIs",
	"confidence": 1,
	"top": 664,
	"left": 453,
	"width": 108,
	"height": 30,
	"Info": [{
		"property": "finegrained_vertexes_location",
		"value": "[{\"x\":453,\"y\":664},{\"x\":476,\"y\":664},
				 ...,{\"x\":453,\"y\":669}]" // 16 points
	}, {
		"property": "min_finegrained_vertexes_location",
		"value": "[{\"x\":453,\"y\":664},{\"x\":559,\"y\":664},
				 {\"x\":559,\"y\":692},{\"x\":453,\"y\":692}]"
	}]
}]
GeneXusAI - Baidu - OCR
 4335ms
Google 
[{
    "label": "Levi'S",
    "confidence": 1,
    "top": 821,
    "left": 535,
    "width": 139,
    "height": 40,
    "Info": [{
            "property": "Language",
            "value": "en"
        }
    ]
}]
GeneXusAI - Google - OCR
 8305ms
IBM 
[{
    "label": "vrs",
    "confidence": 0.697,
    "top": 241,
    "left": 153,
    "width": 8,
    "height": 25,
    "Info": [{
        "property": "line_number",
        "value": "0.00000"
    }]
}, {
    "label": "le",
    "confidence": 0.668,
    "top": 247,
    "left": 162,
    "width": 35,
    "height": 14,
    "Info": [{
        "property": "line_number",
        "value": "0.00000"
    }]
}]
GeneXusAI - IBM - OCR
 9251ms
Microsoft 
[{
    "label": "Levrs",
    "confidence": 0.0,
    "top": 823,
    "left": 538,
    "width": 147,
    "height": 36,
    "Info": [{
        "property": "Angle",
        "value": "0.00000"
    }, {
        "property": "Language",
        "value": "tr"
    }, {
        "property": "Orientation",
        "value": "Up"
    }]
}]
GeneXusAI - Microsofot - OCR
 8496ms
MLKit 
[{
    "label": "Levrs",
    "confidence": 1,
    "top": 806,
    "left": 516,
    "width": 200,
    "height": 75
}]
GeneXusAI - Microsofot - OCR
 978ms
SAP 
[{
    "label": "Levis",
    "confidence": 0.955528974533081,
    "top": 820,
    "left": 534,
    "width": 156,
    "height": 39
}]
GeneXusAI - SAP - OCR
 8595ms
Tencent 
[{
    "label": "LeVi's",
    "confidence": 0.795,
    "top": 668,
    "left": 436,
    "width": 121,
    "height": 28
}]
GeneXusAI - Tencent - OCR
 7582ms

Another example, when the input image has clean text (reduced noise).

image_201887102737_1_png

Provider Output Benchmark
Alibaba 
[{
	"label": "GeneXus",
	"confidence": 0.999,
	"top": 52,
	"left": 20,
	"width": 63,
	"height": 24
}, {
	"label": "LEGACY",
	"confidence": 0.999,
	"top": 70,
	"left": 355,
	"width": 107,
	"height": 25
}, {
	"label": "MODERNIZATION",
	"confidence": 0.999,
	"top": 103,
	"left": 356,
	"width": 230,
	"height": 25
}, {
	"label": "WITH GENEXUSTM15",
	"confidence": 0.999,
	"top": 138,
	"left": 357,
	"width": 204,
	"height": 22
}, {
	"label": "DOWNLOAD THE EBOOK!",
	"confidence": 0.999,
	"top": 228,
	"left": 378,
	"width": 149,
	"height": 14
}]
GeneXusAI - Alibaba - OCR - Clean
 4171ms
Amazon 
[{
    "label": "LEGACY",
    "confidence": 0.999,
    "top": 103,
    "left": 546,
    "width": 161,
    "height": 41
}, {
    "label": "MODERNIZATION",
    "confidence": 0.999,
    "top": 158,
    "left": 550,
    "width": 340,
    "height": 38
}, {
    "label": "WITH",
    "confidence": 0.999,
    "top": 211,
    "left": 546,
    "width": 86,
    "height": 31
}, {
    "label": "15",
    "confidence": 0.998,
    "top": 211,
    "left": 821,
    "width": 42,
    "height": 35
}, {
    "label": "EBOOK!",
    "confidence": 0.997,
    "top": 350,
    "left": 736,
    "width": 77,
    "height": 20
}, {
    "label": "THE",
    "confidence": 0.996,
    "top": 352,
    "left": 695,
    "width": 40,
    "height": 20
}, {
    "label": "ATES",
    "confidence": 0.992,
    "top": 289,
    "left": 364,
    "width": 42,
    "height": 17
}, {
    "label": "GENEXUSTN",
    "confidence": 0.983,
    "top": 211,
    "left": 638,
    "width": 167,
    "height": 34
}, {
    "label": "DOWNLOAD",
    "confidence": 0.973,
    "top": 350,
    "left": 580,
    "width": 113,
    "height": 20
}, {
    "label": "illsion",
    "confidence": 0.945,
    "top": 95,
    "left": 270,
    "width": 32,
    "height": 11
}, {
    "label": "3LD",
    "confidence": 0.944,
    "top": 308,
    "left": 362,
    "width": 29,
    "height": 15
}, {
    "label": "3ut it is!",
    "confidence": 0.938,
    "top": 134,
    "left": 263,
    "width": 32,
    "height": 8
}, {
    "label": "is",
    "confidence": 0.933,
    "top": 112,
    "left": 293,
    "width": 9,
    "height": 7
}, {
    "label": "Genekus:",
    "confidence": 0.872,
    "top": 92,
    "left": 28,
    "width": 101,
    "height": 24
}, {
    "label": "'condinve",
    "confidence": 0.863,
    "top": 121,
    "left": 261,
    "width": 38,
    "height": 10
}, {
    "label": "ian",
    "confidence": 0.837,
    "top": 98,
    "left": 257,
    "width": 16,
    "height": 7
}]
GeneXusAI - Amazon - OCR - Clean
 2418ms
Baidu 
[{
	"label": "Genexus",
	"confidence": 1.000,
	"top": 83,
	"left": 30,
	"width": 97,
	"height": 34,
	"Info": [{
		"property": "finegrained_vertexes_location",
		"value": "[{\"x\":30,\"y\":83},{\"x\":56,\"y\":83},
				   ...,{\"x\":30,\"y\":90}]" // 38 points
	}, {
		"property": "min_finegrained_vertexes_location",
		"value": "[{\"x\":30,\"y\":83},{\"x\":126,\"y\":83},
				   {\"x\":126,\"y\":116},{\"x\":30,\"y\":116}]"
	}]
}, {
	"label": "uE it is",
	"confidence": 1.000,
	"top": 134,
	"left": 267,
	"width": 45,
	"height": 11,
	"Info": [{
		"property": "finegrained_vertexes_location",
		"value": "[{\"x\":267,\"y\":134},{\"x\":273,\"y\":134},
				   ...,{\"x\":267,\"y\":135}]" // 38 points
	}, {
		"property": "min_finegrained_vertexes_location",
		"value": "[{\"x\":267,\"y\":134},{\"x\":311,\"y\":134},
				   {\"x\":311,\"y\":144},{\"x\":267,\"y\":144}]"
	}]
}, {
	"label": "EGACY",
	"confidence": 1.000,
	"top": 107,
	"left": 571,
	"width": 143,
	"height": 42,
	"Info": [{
			"property": "finegrained_vertexes_location",
			"value": "[{"x":571,"y":107},{"x":602,"y":107},
					   ...,{"x":571,"y":116}]" // 38 points
		}, {
			"property": "min_finegrained_vertexes_location",
			"value": "[{\"x\":570,\"y\":107},{\"x\":711,\"y\":107},
			           {\"x\":711,\"y\":147},{\"x\":570,\"y\":147}]"
		}
	]
}, {
	"label": "MODERNIZATION",
	"confidence": 1.000,
	"top": 157,
	"left": 545,
	"width": 363,
	"height": 44,
	"Info": [{
		"property": "finegrained_vertexes_location",
		"value": "[{\"x\":545,\"y\":157},{\"x\":577,\"y\":157},
				  ...,{\"x\":545,\"y\":168}]" // 38 points
	}, {
		"property": "min_finegrained_vertexes_location",
		"value": "[{\"x\":545,\"y\":157},{\"x\":906,\"y\":157},
				   {\"x\":906,\"y\":199},{\"x\":545,\"y\":199}]"
	}]
}, {
	"label": "WITH GENEXUSM 15",
	"confidence": 1.000,
	"top": 213,
	"left": 545,
	"width": 322,
	"height": 35,
	"Info": [{
		"property": "finegrained_vertexes_location",
		"value": "[{\"x\":545,\"y\":213},{\"x\":571,\"y\":213},
		          ...,{\"x\":545,\"y\":220}]" // 38 points
	}, {
		"property": "min_finegrained_vertexes_location",
		"value": "[{\"x\":545,\"y\":213},{\"x\":865,\"y\":213},
				   {\"x\":865,\"y\":247},{\"x\":545,\"y\":247}]"
	}]
}, {
	"label": "DOWNLOAD THE EBOOK!",
	"confidence": 1.000,
	"top": 354,
	"left": 583,
	"width": 232,
	"height": 21,
	"Info": [{
		"property": "finegrained_vertexes_location",
		"value": "[{\"x\": 583,\"y\": 354},{\"x\": 598,\"y\": 354},
				  ...,{\"x\": 583,\"y\": 359}]" // 38 points
	}, {
		"property": "min_finegrained_vertexes_location",
		"value": [{\"x\":582,\"y\":353},{\"x\":812,\"y\":353},
		          {\"x\":812,\"y\":373},{\"x\":582,\"y\":373}]"
	}]
}]
GeneXusAI - Baidu - OCR - Clean
 2418ms
Google 
[{
    "label": "LEGACY",
    "confidence": 1,
    "top": 110,
    "left": 551,
    "width": 155,
    "height": 34,
    "Info": [{
            "property": "Language",
            "value": ""
        }
    ]
}, {
    "label": "MODERNIZATION",
    "confidence": 1,
    "top": 162,
    "left": 553,
    "width": 347,
    "height": 35,
    "Info": [{
            "property": "Language",
            "value": ""
        }
    ]
}, {
    "label": "WITH",
    "confidence": 1,
    "top": 216,
    "left": 553,
    "width": 77,
    "height": 29,
    "Info": [{
            "property": "Language",
            "value": ""
        }
    ]
}, {
    "label": "GENEXUS",
    "confidence": 1,
    "top": 216,
    "left": 641,
    "width": 150,
    "height": 29,
    "Info": [{
            "property": "Language",
            "value": ""
        }
    ]
}, {
    "label": "15",
    "confidence": 1,
    "top": 216,
    "left": 823,
    "width": 31,
    "height": 29,
    "Info": [{
            "property": "Language",
            "value": ""
        }
    ]
}, {
    "label": "Mlusion",
    "confidence": 1,
    "top": 99,
    "left": 274,
    "width": 24,
    "height": 7,
    "Info": [{
            "property": "Language",
            "value": ""
        }
    ]
}, {
    "label": "GeneXus",
    "confidence": 1,
    "top": 96,
    "left": 34,
    "width": 85,
    "height": 3,
    "Info": [{
            "property": "Language",
            "value": ""
        }
    ]
}, {
    "label": "it",
    "confidence": 1,
    "top": 136,
    "left": 279,
    "width": 4,
    "height": 5,
    "Info": [{
            "property": "Language",
            "value": ""
        }
    ]
}, {
    "label": "is!",
    "confidence": 1,
    "top": 136,
    "left": 287,
    "width": 8,
    "height": 7,
    "Info": [{
            "property": "Language",
            "value": ""
        }
    ]
}, {
    "label": "(0S",
    "confidence": 1,
    "top": 262,
    "left": 291,
    "width": 9,
    "height": 7,
    "Info": [{
            "property": "Language",
            "value": ""
        }
    ]
}, {
    "label": "430",
    "confidence": 1,
    "top": 263,
    "left": 304,
    "width": 9,
    "height": 7,
    "Info": [{
            "property": "Language",
            "value": ""
        }
    ]
}, {
    "label": "ees",
    "confidence": 1,
    "top": 263,
    "left": 318,
    "width": 16,
    "height": 8,
    "Info": [{
            "property": "Language",
            "value": ""
        }
    ]
}, {
    "label": "ATES",
    "confidence": 1,
    "top": 295,
    "left": 369,
    "width": 31,
    "height": 18,
    "Info": [{
            "property": "Language",
            "value": ""
        }
    ]
}, {
    "label": "RLD",
    "confidence": 1,
    "top": 313,
    "left": 367,
    "width": 20,
    "height": 13,
    "Info": [{
            "property": "Language",
            "value": ""
        }
    ]
}, {
    "label": "DOWNLOAD",
    "confidence": 1,
    "top": 356,
    "left": 586,
    "width": 106,
    "height": 15,
    "Info": [{
            "property": "Language",
            "value": ""
        }
    ]
}, {
    "label": "THE",
    "confidence": 1,
    "top": 356,
    "left": 698,
    "width": 36,
    "height": 15,
    "Info": [{
            "property": "Language",
            "value": ""
        }
    ]
}, {
    "label": "EBOOK!",
    "confidence": 1,
    "top": 356,
    "left": 741,
    "width": 69,
    "height": 15,
    "Info": [{
            "property": "Language",
            "value": ""
        }
    ]
}]
GeneXusAI - Google - OCR - Clean
 2716ms
IBM 
[{
    "label": "gene",
    "confidence": 0.7228,
    "top": 92,
    "left": 35,
    "width": 62,
    "height": 27,
    "Info": [{
        "property": "line_number",
        "value": "0.00000"
    }]
}, {
    "label": "nexus",
    "confidence": 0.8237,
    "top": 80,
    "left": 62,
    "width": 68,
    "height": 32,
    "Info": [{
        "property": "line_number",
        "value": "0.00000"
    }]
}, {
    "label": "legacy",
    "confidence": 0.9684,
    "top": 111,
    "left": 553,
    "width": 158,
    "height": 37,
    "Info": [{
        "property": "line_number",
        "value": "1.00000"
    }]
}, {
    "label": "modernization",
    "confidence": 0.9898,
    "top": 163,
    "left": 553,
    "width": 351,
    "height": 37,
    "Info": [{
        "property": "line_number",
        "value": "2.00000"
    }]
}, 
...
}]
GeneXusAI - IBM - OCR - Clean
 12898ms
Microsoft 
[{
    "label": "GeneXus",
    "confidence": 0,
    "top": 94,
    "left": 0,
    "width": 126,
    "height": 76,
    "Info": [{
            "property": "Angle",
            "value": "0.00000"
        }, {
            "property": "Language",
            "value": "en"
        }, {
            "property": "Orientation",
            "value": "up"
        }
    ]
}, {
    "label": "WITH GENEXUSTM 15",
    "confidence": 0,
    "top": 216,
    "left": 0,
    "width": 862,
    "height": 215,
    "Info": [{
            "property": "Angle",
            "value": "0.00000"
        }, {
            "property": "Language",
            "value": "en"
        }, {
            "property": "Orientation",
            "value": "up"
        }
    ]
}, {
    "label": "DOWNLOAD THE EBOOK!",
    "confidence": 0,
    "top": 353,
    "left": 0,
    "width": 813,
    "height": 352,
    "Info": [{
            "property": "Angle",
            "value": "0.00000"
        }, {
            "property": "Language",
            "value": ""
        }, {
            "property": "Orientation",
            "value": ""
        }
    ]
},
...
]
GeneXusAI - MCS - OCR - Clean
 8496ms
MLKit 
[]
N/A
 -ms
SAP 
[{
	"label": "MODERNIZATION",
	"confidence": 0.995006024837494,
	"top": 159,
	"left": 562,
	"width": 322,
	"height": 41
}, {
	"label": "LEGACY",
	"confidence": 0.953990995883942,
	"top": 109,
	"left": 551,
	"width": 161,
	"height": 37
}, {
	"label": "Genexus",
	"confidence": 0.974162995815277,
	"top": 79,
	"left": 30,
	"width": 99,
	"height": 40
}, {
	"label": "DOWNLOAD",
	"confidence": 0.931997001171112,
	"top": 355,
	"left": 586,
	"width": 106,
	"height": 18
}, {
	"label": "GENEXUS",
	"confidence": 0.976303994655609,
	"top": 216,
	"left": 641,
	"width": 155,
	"height": 30
}, {
	"label": "WITH",
	"confidence": 0.947585999965668,
	"top": 217,
	"left": 550,
	"width": 83,
	"height": 27
}, {
	"label": "RATES",
	"confidence": 0.932789027690888,
	"top": 294,
	"left": 364,
	"width": 39,
	"height": 19
}, {
	"label": "EBOOK!",
	"confidence": 0.912930011749268,
	"top": 354,
	"left": 742,
	"width": 71,
	"height": 19
}, {
	"label": "RLD",
	"confidence": 0.899074018001556,
	"top": 310,
	"left": 362,
	"width": 28,
	"height": 18
}, {
	"label": "THE",
	"confidence": 0.899158000946045,
	"top": 356,
	"left": 698,
	"width": 39,
	"height": 16
}, {
	"label": "15",
	"confidence": 0.925707995891571,
	"top": 217,
	"left": 826,
	"width": 36,
	"height": 29
}]
GeneXusAI - SAP - OCR - Clean
 3780ms
Tencent 
[{
    "label": "15",
    "confidence": 0.99,
    "top": 232,
    "left": 856,
    "width": 37,
    "height": 34
}, {
    "label": "MODE",
    "confidence": 0.975,
    "top": 242,
    "left": 577,
    "width": 96,
    "height": 70
}, {
    "label": "RLD",
    "confidence": 0.96,
    "top": 480,
    "left": 452,
    "width": 22,
    "height": 13
}, {
    "label": "RATES",
    "confidence": 0.955,
    "top": 461,
    "left": 447,
    "width": 36,
    "height": 17
}, {
    "label": "GENF",
    "confidence": 0.943,
    "top": 277,
    "left": 685,
    "width": 63,
    "height": 46
}, {
    "label": "THEE",
    "confidence": 0.873,
    "top": 396,
    "left": 780,
    "width": 38,
    "height": 44
}, {
    "label": "ERi",
    "confidence": 0.868,
    "top": 229,
    "left": 675,
    "width": 52,
    "height": 50
}, {
    "label": "NEXUS\"",
    "confidence": 0.863,
    "top": 242,
    "left": 746,
    "width": 89,
    "height": 68
}, {
    "label": "NZATON",
    "confidence": 0.848,
    "top": 159,
    "left": 735,
    "width": 180,
    "height": 103
}, {
    "label": "LEGAER",
    "confidence": 0.828,
    "top": 174,
    "left": 559,
    "width": 149,
    "height": 94
}, {
    "label": "?",
    "confidence": 0.775,
    "top": 350,
    "left": 120,
    "width": 29,
    "height": 42
}, {
    "label": "Gene",
    "confidence": 0.775,
    "top": 369,
    "left": 68,
    "width": 46,
    "height": 37
}, {
    "label": "EBDON",
    "confidence": 0.759,
    "top": 372,
    "left": 823,
    "width": 67,
    "height": 46
}, {
    "label": "MITHC",
    "confidence": 0.712,
    "top": 300,
    "left": 591,
    "width": 87,
    "height": 59
}, {
    "label": "auters",
    "confidence": 0.71,
    "top": 318,
    "left": 288,
    "width": 33,
    "height": 12
}, {
    "label": "OMCOO",
    "confidence": 0.614,
    "top": 411,
    "left": 677,
    "width": 97,
    "height": 58
}, {
    "label": "HhIsion",
    "confidence": 0.602,
    "top": 308,
    "left": 294,
    "width": 27,
    "height": 9
}]
GeneXusAI - Tencent - OCR - Clean
 1827ms

Notes

  • The label assigned for an object depends on the provider used (i.e. string detection accuracy). Additional information can be found on the OutputRegion.Info field if it is given by the provider.
  • Some providers are more accurate with noisy images than others (e.g. stamps on shirts). All of them work fine with clean text on the image (e.g. a scanned text).
  • Maximum image file size is 10MB.
  • GeneXusAI does not provide support for drawing a rectangle over an image. This action is the responsibility of the developer.
    TIP: For Web applications, a good alternative can be combining HTML5 Canvas control with JavaScript with User Control object. On the other hand, for Smart Devices you could use Image Map control on which you can set the processed image as background and 'draw' square regions (i.e. set a border color on the table item of the grid).
  • IBM provider: Deprecated as of September 12 (2019). Check IBM Visual Recognition's Release Notes. Before deprecation, GeneXusAI sses the dark beta feature of Visual Recognition and you had to request access.

Scope

Platforms Web(.NET,.NETCore,Java), SmartDevices(Android,iOS)
Connectivity Online

Availability

This procedure is available as of GeneXus 16.