Updated Rekognition with Text Detection

CFDinesh · Aug 6, 2018 · 55cae67 · 55cae67
1 parent bc6f054
commit 55cae67
Show file tree

Hide file tree

Showing 2 changed files with 79 additions and 3 deletions.
diff --git a/model/rekognitionLib.cfc b/model/rekognitionLib.cfc
@@ -94,7 +94,53 @@ component output="false" hint="A utility for making requests to AWS Rekognition.
 	}
 
 	/**
-	*	@description Returns a structure of labels for the image, with the label as the key and the confidence level as the value
+	*	@description Returns an structure of both the lines of text found in the image and the individual words
+	*	@requiredArguments
+	*		- awsBucketName = Name of the bucket where the images reside
+	*		- pathToImage = Path to the image in the provided bucket
+	*/
+	public struct function detectText(required string awsBucketName, required string pathToImage) {
+		var returnStruct = structNew();
+		var thisDetectionObj = 0;
+		var linesCounter = 0;
+		var wordsCounter = 0;
+		var detectTextRequest = CreateObject('java', 'com.amazonaws.services.rekognition.model.DetectTextRequest').init();
+		var imageToScan = CreateObject('java', 'com.amazonaws.services.rekognition.model.Image').init();
+		var imageS3Object = CreateObject('java', 'com.amazonaws.services.rekognition.model.S3Object').init();
+		imageS3Object.setBucket(arguments.awsBucketName);
+		imageS3Object.setName(arguments.pathToImage);
+		imageToScan.setS3Object(imageS3Object);
+		detectTextRequest.setImage(imageToScan);
+
+		var detectTextResult = variables.rekognitionService.detectText(detectTextRequest);
+		var detectionsArray = detectTextResult.getTextDetections();
+
+		returnStruct.lines = arrayNew(1);
+		returnStruct.words = arrayNew(1);
+
+		detectionsArray.each(function(thisDetectionObj, index) {
+			if (thisDetectionObj.getType() is "LINE") {
+				linesCounter++;
+				returnStruct.lines[linesCounter] = structNew();
+				returnStruct.lines[linesCounter]['label'] = thisDetectionObj.getDetectedText();
+				returnStruct.lines[linesCounter]['confidence'] = Int(thisDetectionObj.getConfidence());
+				returnStruct.lines[linesCounter]['id'] = thisDetectionObj.getID();
+				returnStruct.lines[linesCounter]['geometry'] = thisDetectionObj.getGeometry().toString();
+			} else {
+				wordsCounter++;
+				returnStruct.words[wordsCounter] = structNew();
+				returnStruct.words[wordsCounter]['label'] = thisDetectionObj.getDetectedText();
+				returnStruct.words[wordsCounter]['confidence'] = Int(thisDetectionObj.getConfidence());
+				returnStruct.words[wordsCounter]['id'] = thisDetectionObj.getID();
+				returnStruct.words[wordsCounter]['parentID'] = thisDetectionObj.getParentID();
+				returnStruct.words[wordsCounter]['geometry'] = thisDetectionObj.getGeometry().toString();
+			}
+		});
+		return returnStruct;
+	}
+
+	/**
+	*	@description Returns an array of labels for the image as structures, with the label as the key and the confidence level as the value
 	*	@requiredArguments
 	*		- awsBucketName = Name of the bucket where the images reside
 	*		- pathToImage = Path to the image in the provided bucket

diff --git a/rekognition.cfm b/rekognition.cfm
@@ -3,8 +3,9 @@
 <!--- Put the URLs to your images in S3 here --->
 <cfset s3images = structNew()>
 <cfset s3images.awsBucketName = "YOUR BUCKET NAME GOES HERE" />
-<cfset s3images.facesForMatching = [ "ARRAY","OF","PATHS","TO","IMAGES","IN","THE","BUCKET","LISTED","ABOVE"] />
-<cfset s3images.imagesForLabels = [ "ARRAY","OF","PATHS","TO","IMAGES","IN","THE","BUCKET","LISTED","ABOVE] />
+<cfset s3images.facesForMatching = [ "ARRAY","OF","PATHS","TO","IMAGES","IN","THE","BUCKET","LISTED","ABOVE" ] />
+<cfset s3images.imagesForLabels = [ "ARRAY","OF","PATHS","TO","IMAGES","IN","THE","BUCKET","LISTED","ABOVE" ] />
+<cfset s3images.imagesWithText = [ "ARRAY","OF","PATHS","TO","IMAGES","IN","THE","BUCKET","LISTED","ABOVE" ] />
 
 <cfif structKeyExists(URL, "rekogRequest")>
 	<cfscript>
@@ -32,6 +33,11 @@
 				detectSentimentResult = rekognitionLib.detectSentiment(s3images.awsBucketName, sourceImage);
 				break;
 
+			case 'detectText':
+				sourceImage = s3images.imagesWithText[randRange(1, arrayLen(s3images.imagesWithText))];
+				detectTextResult = rekognitionLib.detectText(s3images.awsBucketName, sourceImage);
+				break;
+
 			default:
 				throw(message="Unsupported service requested", detail="You have requested a method (#URL.rekogRequest#) which is not supported at this time.");
 				break;
@@ -108,12 +114,36 @@
 								<br clear="all">
 							</cfoutput>
 						</cfcase>
+						<cfcase value="detectText">
+							<cfoutput>
+								<div>
+									<div style="width:50%; float:left;">
+										<p>Here is the image used:</p>
+										<p>
+											<img src="http://#s3images.awsBucketName#.s3.amazonaws.com/#sourceImage#" width="450" height="350" border="1" />
+										</p>
+									</div>
+									<div style="width:50%; float:right;">
+										<p>Lines of text:</p>
+										<cfloop array="#detectTextResult.lines#" index="idxThisLine">
+											#idxThisLine.id# | #idxThisLine.label# | (#idxThisLine.confidence#%)<br/>
+										</cfloop>
+										<p>Individual words:</p>
+										<cfloop array="#detectTextResult.words#" index="idxThisWord">
+											Line: #idxThisWord.parentID# &mdash; #idxThisWord.label# (#idxThisWord.confidence#%)<br/>
+										</cfloop>
+									</div>
+								</div>
+								<br clear="all">
+							</cfoutput>
+						</cfcase>
 					</cfswitch>
 				</cfif> <!--- End if a Rekognition request was made --->
 
 				<p><a href="rekognition.cfm?rekogRequest=compareFaces">Compare Two Faces</a></p>
 				<p><a href="rekognition.cfm?rekogRequest=detectLabels">Label the Properties of an Image</a></p>
 				<p><a href="rekognition.cfm?rekogRequest=detectSentiment">Detect Facial Sentiment of an Image</a></p>
+				<p><a href="rekognition.cfm?rekogRequest=detectText">Detect Text in an Image</a></p>
 
 				<p align="right" ><a href="index.cfm" class="homeButton">Home</a></p>
 			</div>