SpringCloud/spring-functions-catalog
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

GH-7: Add checkstyle and javaformat plugins

Browse Source 
Fixes: #7

* Run `./gradlew format`
* Updates from PR review suggestions
This commit is contained in:
Chris Bono
2023-12-29 21:08:58 -06:00
committed by Artem Bilan
parent 84e732da08
commit 836708f0f2
357 changed files with 4344 additions and 4519 deletions
Download Patch File Download Diff File Expand all files Collapse all files

187
function/spring-image-recognition-function/src/main/java/org/springframework/cloud/fn/image/recognition/ImageRecognition.java
View File

@@ -46,23 +46,29 @@ import org.springframework.util.StreamUtils;
public class ImageRecognition implements AutoCloseable {
private final List<String> labels;
private final GraphRunner imageNormalization;
private final GraphRunner imageRecognition;
private final GraphRunner maxProbability;
private final GraphRunner topKProbabilities;
/**
* Instead of creating the {@link ImageRecognition} service explicitly via the constructor,
* you should consider the convenience factory methods below. E.g.
*
* {@link #inception(String, int, int, boolean)}
* {@link #mobileNetV1(String, int, int, boolean)}
* {@link #mobileNetV2(String, int, int, boolean)}
* Instead of creating the {@link ImageRecognition} service explicitly via the
* constructor, you should consider the convenience factory methods below. E.g.
*
* {@link #inception(String, int, int, boolean)}
* {@link #mobileNetV1(String, int, int, boolean)}
* {@link #mobileNetV2(String, int, int, boolean)}
* @param modelUri location of the pre-trained model to use.
* @param labelsUri location of the list fromMemory pre-trained categories used by the model.
* @param imageRecognitionGraphInputName name of the Model's input node to send the input image to.
* @param imageRecognitionGraphOutputName name of the Model's output node to retrieve the predictions from.
* @param labelsUri location of the list fromMemory pre-trained categories used by the
* model.
* @param imageRecognitionGraphInputName name of the Model's input node to send the
* input image to.
* @param imageRecognitionGraphOutputName name of the Model's output node to retrieve
* the predictions from.
* @param imageHeight normalized image height.
* @param imageWidth normalized image width.
* @param mean mean value to normalize the input image.
@@ -71,62 +77,66 @@ public class ImageRecognition implements AutoCloseable {
* @param cacheModel if true the pre-trained model is cached on the local file system.
*/
public ImageRecognition(String modelUri, String labelsUri, int imageHeight, int imageWidth, float mean, float scale,
String imageRecognitionGraphInputName, String imageRecognitionGraphOutputName, int responseSize, boolean cacheModel) {
String imageRecognitionGraphInputName, String imageRecognitionGraphOutputName, int responseSize,
boolean cacheModel) {
this.labels = labels(labelsUri);
/**
* Normalizes the raw input image into format expected by the pre-trained Inception/MobileNetV1/MobileNetV2 models.
* Typically the model is trained fromMemory images scaled to certain size. Usually it is 224x224 pixels, but can be
* also 192x192, 160x160, 128128, 92x92. Use the (imageHeight, imageWidth) to set the desired size.
* The colors, represented as R, G, B in 1-byte each were converted to float using (Value - Mean)/Scale.
* Normalizes the raw input image into format expected by the pre-trained
* Inception/MobileNetV1/MobileNetV2 models. Typically the model is trained
* fromMemory images scaled to certain size. Usually it is 224x224 pixels, but can
* be also 192x192, 160x160, 128128, 92x92. Use the (imageHeight, imageWidth) to
* set the desired size. The colors, represented as R, G, B in 1-byte each were
* converted to float using (Value - Mean)/Scale.
*
* imageHeight normalized image height.
* imageWidth normalized image width.
* mean mean value to normalize the input image.
* scale scale to normalize the input image.
* imageHeight normalized image height. imageWidth normalized image width. mean
* mean value to normalize the input image. scale scale to normalize the input
* image.
*/
this.imageNormalization = new GraphRunner("raw_image", "normalized_image")
.withGraphDefinition(tf -> {
Placeholder<String> input = tf.withName("raw_image").placeholder(String.class);
final Operand<Float> decodedImage =
tf.dtypes.cast(tf.image.decodeJpeg(input, DecodeJpeg.channels(3L)), Float.class);
final Operand<Float> resizedImage = tf.image.resizeBilinear(
tf.expandDims(decodedImage, tf.constant(0)),
tf.constant(new int[] { imageHeight, imageWidth }));
tf.withName("normalized_image").math.div(tf.math.sub(resizedImage, tf.constant(mean)), tf.constant(scale));
});
this.imageNormalization = new GraphRunner("raw_image", "normalized_image").withGraphDefinition(tf -> {
Placeholder<String> input = tf.withName("raw_image").placeholder(String.class);
final Operand<Float> decodedImage = tf.dtypes.cast(tf.image.decodeJpeg(input, DecodeJpeg.channels(3L)),
Float.class);
final Operand<Float> resizedImage = tf.image.resizeBilinear(tf.expandDims(decodedImage, tf.constant(0)),
tf.constant(new int[] { imageHeight, imageWidth }));
tf.withName("normalized_image").math.div(tf.math.sub(resizedImage, tf.constant(mean)), tf.constant(scale));
});
this.imageRecognition = new GraphRunner(imageRecognitionGraphInputName, imageRecognitionGraphOutputName)
.withGraphDefinition(new ProtoBufGraphDefinition(toResource(modelUri), cacheModel));
.withGraphDefinition(new ProtoBufGraphDefinition(toResource(modelUri), cacheModel));
this.maxProbability = new GraphRunner(Arrays.asList("recognition_result"), Arrays.asList("category", "probability"))
.withGraphDefinition(tf -> {
Placeholder<Float> input = tf.withName("recognition_result").placeholder(Float.class);
tf.withName("category").math.argMax(input, tf.constant(1));
tf.withName("probability").max(input, tf.constant(1));
});
this.maxProbability = new GraphRunner(Arrays.asList("recognition_result"),
Arrays.asList("category", "probability"))
.withGraphDefinition(tf -> {
Placeholder<Float> input = tf.withName("recognition_result").placeholder(Float.class);
tf.withName("category").math.argMax(input, tf.constant(1));
tf.withName("probability").max(input, tf.constant(1));
});
this.topKProbabilities = new GraphRunner("recognition_result", "topK")
.withGraphDefinition(tf -> {
Placeholder<Float> input = tf.withName("recognition_result").placeholder(Float.class);
tf.withName("topK").nn.topK(input, tf.constant(responseSize), TopK.sorted(true));
});
this.topKProbabilities = new GraphRunner("recognition_result", "topK").withGraphDefinition(tf -> {
Placeholder<Float> input = tf.withName("recognition_result").placeholder(Float.class);
tf.withName("topK").nn.topK(input, tf.constant(responseSize), TopK.sorted(true));
});
}
/**
* Takes an byte encoded image and returns the most probable category recognized in the image along fromMemory its probability.
* Takes an byte encoded image and returns the most probable category recognized in
* the image along fromMemory its probability.
* @param inputImage Byte array encoded image to recognize.
* @return Returns a single map entry containing the names of the recognized categories as key and the confidence as value.
* @return Returns a single map entry containing the names of the recognized
* categories as key and the confidence as value.
*/
public Map<String, Double> recognizeMax(byte[] inputImage) {
try (Tensor inputTensor = Tensor.create(inputImage); GraphRunnerMemory memorize = new GraphRunnerMemory()) {
Map<String, Tensor<?>> max = this.imageNormalization.andThen(memorize)
.andThen(this.imageRecognition).andThen(memorize)
.andThen(this.maxProbability).andThen(memorize)
.apply(Collections.singletonMap("raw_image", inputTensor));
.andThen(this.imageRecognition)
.andThen(memorize)
.andThen(this.maxProbability)
.andThen(memorize)
.apply(Collections.singletonMap("raw_image", inputTensor));
long[] category = new long[1];
max.get("category").copyTo(category);
@@ -138,26 +148,28 @@ public class ImageRecognition implements AutoCloseable {
}
/**
* Takes an byte encoded input image and returns the top K most probable categories recognized in the image
* along fromMemory their probabilities.
*
* Takes an byte encoded input image and returns the top K most probable categories
* recognized in the image along fromMemory their probabilities.
* @param inputImage Byte array encoded image to recognize.
* @return Returns a list of key-value pairs. Every key-value pair represents a single category recognized.
* The key stands for the name(s) of the category while the value states the confidence that there is an
* object of this category. The entries in the Map are ordered from the higher to the lower confidences.
* @return Returns a list of key-value pairs. Every key-value pair represents a single
* category recognized. The key stands for the name(s) of the category while the value
* states the confidence that there is an object of this category. The entries in the
* Map are ordered from the higher to the lower confidences.
*/
public Map<String, Double> recognizeTopK(byte[] inputImage) {
try (Tensor inputTensor = Tensor.create(inputImage); GraphRunnerMemory memorize = new GraphRunnerMemory()) {
Map<String, Tensor<?>> topKResults =
this.imageNormalization.andThen(memorize)
.andThen(this.imageRecognition).andThen(memorize)
.andThen(this.topKProbabilities).andThen(memorize)
.apply(Collections.singletonMap("raw_image", inputTensor));
Map<String, Tensor<?>> topKResults = this.imageNormalization.andThen(memorize)
.andThen(this.imageRecognition)
.andThen(memorize)
.andThen(this.topKProbabilities)
.andThen(memorize)
.apply(Collections.singletonMap("raw_image", inputTensor));
Tensor recognizedImagesTensor = memorize.getTensorMap().get(this.imageRecognition.getSingleFetchName());
float[][] results = new float[(int) recognizedImagesTensor.shape()[0]][(int) recognizedImagesTensor.shape()[1]];
float[][] results = new float[(int) recognizedImagesTensor.shape()[0]][(int) recognizedImagesTensor
.shape()[1]];
recognizedImagesTensor.copyTo(results);
Tensor<Float> topKTensor = topKResults.get("topK").expect(Float.class);
@@ -204,12 +216,10 @@ public class ImageRecognition implements AutoCloseable {
* The Inception graph uses "input" as input and "output" as output.
*
*/
public static ImageRecognition inception(String inceptionModelUri,
int normalizedImageSize, int responseSize, boolean cacheModel) {
return new ImageRecognition(inceptionModelUri, "classpath:/labels/inception_labels.txt",
normalizedImageSize, normalizedImageSize, 117f, 1f,
"input", "output",
responseSize, cacheModel);
public static ImageRecognition inception(String inceptionModelUri, int normalizedImageSize, int responseSize,
boolean cacheModel) {
return new ImageRecognition(inceptionModelUri, "classpath:/labels/inception_labels.txt", normalizedImageSize,
normalizedImageSize, 117f, 1f, "input", "output", responseSize, cacheModel);
}
/**
@@ -218,48 +228,50 @@ public class ImageRecognition implements AutoCloseable {
*
* The normalized image size is always square (e.g. H=W)
*
* The MobileNetV2 graph uses "input" as input and "MobilenetV2/Predictions/Reshape_1" as output.
*
* The MobileNetV2 graph uses "input" as input and "MobilenetV2/Predictions/Reshape_1"
* as output.
* @param mobileNetV2ModelUri model uri
* @param normalizedImageSize Depends on the pre-trained model used. Usually 224px is used.
* @param normalizedImageSize Depends on the pre-trained model used. Usually 224px is
* used.
* @param responseSize Number of responses fot topK requests.
* @param cacheModel cache model
* @return ImageRecognition instance configured fromMemory a MobileNetV2 pre-trained model.
* @return ImageRecognition instance configured fromMemory a MobileNetV2 pre-trained
* model.
*/
public static ImageRecognition mobileNetV2(String mobileNetV2ModelUri,
int normalizedImageSize, int responseSize, boolean cacheModel) {
return new ImageRecognition(mobileNetV2ModelUri, "classpath:/labels/mobilenet_labels.txt",
normalizedImageSize, normalizedImageSize, 0f, 127f,
"input", "MobilenetV2/Predictions/Reshape_1",
responseSize, cacheModel);
public static ImageRecognition mobileNetV2(String mobileNetV2ModelUri, int normalizedImageSize, int responseSize,
boolean cacheModel) {
return new ImageRecognition(mobileNetV2ModelUri, "classpath:/labels/mobilenet_labels.txt", normalizedImageSize,
normalizedImageSize, 0f, 127f, "input", "MobilenetV2/Predictions/Reshape_1", responseSize, cacheModel);
}
/**
* Convenience for MobileNetV1 pre-trained models:
* https://github.com/tensorflow/models/blob/master/research/slim/nets/mobilenet_v1.md#pre-trained-models
*
* The MobileNetV1 graph uses "input" as input and "MobilenetV1/Predictions/Reshape_1" as output.
* The MobileNetV1 graph uses "input" as input and "MobilenetV1/Predictions/Reshape_1"
* as output.
*
*/
public static ImageRecognition mobileNetV1(String mobileNetV1ModelUri,
int normalizedImageSize, int responseSize, boolean cacheModel) {
return new ImageRecognition(mobileNetV1ModelUri, "classpath:/labels/mobilenet_labels.txt",
normalizedImageSize, normalizedImageSize,
0f, 127f,
"input", "MobilenetV1/Predictions/Reshape_1",
responseSize, cacheModel);
public static ImageRecognition mobileNetV1(String mobileNetV1ModelUri, int normalizedImageSize, int responseSize,
boolean cacheModel) {
return new ImageRecognition(mobileNetV1ModelUri, "classpath:/labels/mobilenet_labels.txt", normalizedImageSize,
normalizedImageSize, 0f, 127f, "input", "MobilenetV1/Predictions/Reshape_1", responseSize, cacheModel);
}
/**
* Convert image recognition results into {@link RecognitionResponse} domain list.
* @param recognitionMap map containing the category mames and its probability. Returned by the
* {@link ImageRecognition#recognizeMax(byte[])} and the ImageRecognition{@link #recognizeTopK(byte[])} methods
* @return List of {@link RecognitionResponse} objects representing the name-to-probability pairs in the input map.
* @param recognitionMap map containing the category mames and its probability.
* Returned by the {@link ImageRecognition#recognizeMax(byte[])} and the
* ImageRecognition{@link #recognizeTopK(byte[])} methods
* @return List of {@link RecognitionResponse} objects representing the
* name-to-probability pairs in the input map.
*/
public static List<RecognitionResponse> toRecognitionResponse(Map<String, Double> recognitionMap) {
return recognitionMap.entrySet().stream()
.map(nameProbabilityPair -> new RecognitionResponse(nameProbabilityPair.getKey(), nameProbabilityPair.getValue()))
.collect(Collectors.toList());
return recognitionMap.entrySet()
.stream()
.map(nameProbabilityPair -> new RecognitionResponse(nameProbabilityPair.getKey(),
nameProbabilityPair.getValue()))
.collect(Collectors.toList());
}
@Override
@@ -269,4 +281,5 @@ public class ImageRecognition implements AutoCloseable {
this.maxProbability.close();
this.topKProbabilities.close();
}
}

3
function/spring-image-recognition-function/src/main/java/org/springframework/cloud/fn/image/recognition/ImageRecognitionAugmenter.java
View File

@@ -33,7 +33,6 @@ import javax.imageio.ImageIO;
import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory;
/**
* Ability to to augment the input image fromMemory the recognized labels.
*
@@ -47,6 +46,7 @@ public class ImageRecognitionAugmenter implements BiFunction<byte[], List<Recogn
public static final String IMAGE_FORMAT = "jpg";
private final Color textColor = Color.BLACK;
private final Color bgColor = new Color(167, 252, 0);
public ImageRecognitionAugmenter() {
@@ -54,7 +54,6 @@ public class ImageRecognitionAugmenter implements BiFunction<byte[], List<Recogn
/**
* Augment the input image by adding the recognized classes.
*
* @param imageBytes input image as byte array
* @param result computed recognition labels
* @return the image augmented fromMemory recognized labels.

3
function/spring-image-recognition-function/src/main/java/org/springframework/cloud/fn/image/recognition/RecognitionResponse.java
View File

@@ -20,7 +20,9 @@ package org.springframework.cloud.fn.image.recognition;
* @author Christian Tzolov
*/
public class RecognitionResponse {
private String label;
private Double probability;
public RecognitionResponse() {
@@ -51,4 +53,5 @@ public class RecognitionResponse {
public String toString() {
return "{label='" + label + ", probability=" + probability + '}';
}
}

52
function/spring-image-recognition-function/src/main/java/org/springframework/cloud/fn/image/recognition/util/ImageNetReadableNamesWriter.java
View File

@@ -36,22 +36,20 @@ import org.springframework.util.StreamUtils;
/**
* Create a text file mapping label id to human readable string.
*
* Produces a text file where every line represents single category. The line number represents the category id, while
* the line text is human-readable names for the categories fromMemory this imagenet id.
* Produces a text file where every line represents single category. The line number
* represents the category id, while the line text is human-readable names for the
* categories fromMemory this imagenet id.
*
* Based on https://github.com/tensorflow/models/blob/master/research/slim/datasets/imagenet.py#L66
* Based on
* https://github.com/tensorflow/models/blob/master/research/slim/datasets/imagenet.py#L66
*
* We retrieve a synset file, which contains a list of valid synset labels used
* by ILSVRC competition. There is one synset one per line, eg.
* # n01440764
* # n01443537
* We also retrieve a synset_to_human_file, which contains a mapping from synsets
* to human-readable names for every synset in Imagenet. These are stored in a
* tsv format, as follows:
* # n02119247 black fox
* # n02119359 silver fox
* We assign each synset (in alphabetical order) an integer, starting from 1
* (since 0 is reserved for the background class)
* We retrieve a synset file, which contains a list of valid synset labels used by ILSVRC
* competition. There is one synset one per line, eg. # n01440764 # n01443537 We also
* retrieve a synset_to_human_file, which contains a mapping from synsets to
* human-readable names for every synset in Imagenet. These are stored in a tsv format, as
* follows: # n02119247 black fox # n02119359 silver fox We assign each synset (in
* alphabetical order) an integer, starting from 1 (since 0 is reserved for the background
* class)
*
* @author Christian Tzolov
*/
@@ -62,8 +60,10 @@ public final class ImageNetReadableNamesWriter {
/** BASE_URL. */
public final static String BASE_URL = "https://raw.githubusercontent.com/tensorflow/models/master/research/inception/inception/data/";
/** SYNSET_URI. */
public final static String SYNSET_URI = BASE_URL + "imagenet_lsvrc_2015_synsets.txt";
/** SYNSET_TO_HUMAN_URI. */
public final static String SYNSET_TO_HUMAN_URI = BASE_URL + "imagenet_metadata.txt";
@@ -71,19 +71,24 @@ public final class ImageNetReadableNamesWriter {
Charset utf8 = Charset.forName("UTF-8");
try (InputStream synsetIs = toResource(SYNSET_URI).getInputStream();
InputStream synsetToHumanIs = toResource(SYNSET_TO_HUMAN_URI).getInputStream()) {
InputStream synsetToHumanIs = toResource(SYNSET_TO_HUMAN_URI).getInputStream()) {
List<String> synsetList = Arrays.asList(StreamUtils.copyToString(synsetIs, utf8)
.split("\n")).stream().map(l -> l.trim()).collect(Collectors.toList());
List<String> synsetList = Arrays.asList(StreamUtils.copyToString(synsetIs, utf8).split("\n"))
.stream()
.map(l -> l.trim())
.collect(Collectors.toList());
Assert.notNull(synsetList, "Failed to initialize the labels list");
Assert.isTrue(synsetList.size() == 1000, "Labels list is expected to be of " +
"size 1000 but was:" + synsetList.size());
Assert.isTrue(synsetList.size() == 1000,
"Labels list is expected to be of " + "size 1000 but was:" + synsetList.size());
Map<String, String> synsetToHuman = Arrays.asList(StreamUtils.copyToString(synsetToHumanIs, utf8)
.split("\n")).stream().map(s2h -> s2h.split("\t")).collect(Collectors.toMap(s -> s[0], s -> s[1]));
Map<String, String> synsetToHuman = Arrays
.asList(StreamUtils.copyToString(synsetToHumanIs, utf8).split("\n"))
.stream()
.map(s2h -> s2h.split("\t"))
.collect(Collectors.toMap(s -> s[0], s -> s[1]));
Assert.notNull(synsetToHuman, "Failed to initialize the synsetToHuman");
Assert.isTrue(synsetToHuman.size() == 21842, "synsetToHuman is expected to be of " +
"size 21842 but was:" + synsetToHuman.size());
Assert.isTrue(synsetToHuman.size() == 21842,
"synsetToHuman is expected to be of " + "size 21842 but was:" + synsetToHuman.size());
List<String> l = synsetList.stream().map(id -> synsetToHuman.get(id)).collect(Collectors.toList());
@@ -102,4 +107,5 @@ public final class ImageNetReadableNamesWriter {
public static Resource toResource(String uri) {
return new DefaultResourceLoader().getResource(uri);
}
}

47
function/spring-image-recognition-function/src/test/java/org/springframework/cloud/fn/image/recognition/ImageRecognitionExample.java
View File

@@ -42,62 +42,51 @@ public final class ImageRecognitionExample {
// MmobileNetV2 models
// https://github.com/tensorflow/models/tree/master/research/slim/nets/mobilenet#pretrained-models
String mobilenet_v2_modelUri = "https://storage.googleapis.com/mobilenet_v2/checkpoints/mobilenet_v2_1.4_224.tgz#mobilenet_v2_1.4_224_frozen.pb";
//String mobilenet_v2_modelUri = "https://storage.googleapis.com/mobilenet_v2/checkpoints/mobilenet_v2_0.35_96.tgz#mobilenet_v2_0.35_96_frozen.pb";
try (ImageRecognition imageRecognition = ImageRecognition.mobileNetV2(
mobilenet_v2_modelUri,
224,
5,
true)) {
// String mobilenet_v2_modelUri =
// "https://storage.googleapis.com/mobilenet_v2/checkpoints/mobilenet_v2_0.35_96.tgz#mobilenet_v2_0.35_96_frozen.pb";
try (ImageRecognition imageRecognition = ImageRecognition.mobileNetV2(mobilenet_v2_modelUri, 224, 5, true)) {
List<RecognitionResponse> recognizedObjects =
ImageRecognition.toRecognitionResponse(imageRecognition.recognizeTopK(inputImage));
List<RecognitionResponse> recognizedObjects = ImageRecognition
.toRecognitionResponse(imageRecognition.recognizeTopK(inputImage));
// Draw the predicted labels on top of the input image.
byte[] augmentedImage = new ImageRecognitionAugmenter().apply(inputImage, recognizedObjects);
IOUtils.write(augmentedImage, new FileOutputStream("./image-recognition/target/image-augmented-mobilnetV2.jpg"));
IOUtils.write(augmentedImage,
new FileOutputStream("./image-recognition/target/image-augmented-mobilnetV2.jpg"));
String jsonRecognizedObjects = new JsonMapperFunction().apply(recognizedObjects);
System.out.println("mobilnetV2 result:" + jsonRecognizedObjects);
}
String mobilenet_v1_modelUri = "https://download.tensorflow.org/models/mobilenet_v1_2018_08_02/mobilenet_v1_1.0_224.tgz#mobilenet_v1_1.0_224_frozen.pb";
try (ImageRecognition recognitionService = ImageRecognition.mobileNetV1(
mobilenet_v1_modelUri,
224,
5,
true)) {
try (ImageRecognition recognitionService = ImageRecognition.mobileNetV1(mobilenet_v1_modelUri, 224, 5, true)) {
List<RecognitionResponse> recognizedObjects =
ImageRecognition.toRecognitionResponse(recognitionService.recognizeTopK(inputImage));
List<RecognitionResponse> recognizedObjects = ImageRecognition
.toRecognitionResponse(recognitionService.recognizeTopK(inputImage));
// Draw the predicted labels on top of the input image.
byte[] augmentedImage = new ImageRecognitionAugmenter().apply(inputImage, recognizedObjects);
IOUtils.write(augmentedImage, new FileOutputStream("./image-recognition/target/image-augmented-mobilnetV1.jpg"));
IOUtils.write(augmentedImage,
new FileOutputStream("./image-recognition/target/image-augmented-mobilnetV1.jpg"));
String jsonRecognizedObjects = new JsonMapperFunction().apply(recognizedObjects);
System.out.println("mobilnetV1 result:" + jsonRecognizedObjects);
}
String inception_modelUri = "https://storage.googleapis.com/scdf-tensorflow-models/image-recognition/tensorflow_inception_graph.pb";
try (ImageRecognition recognitionService = ImageRecognition.inception(
inception_modelUri,
224,
5,
true)) {
try (ImageRecognition recognitionService = ImageRecognition.inception(inception_modelUri, 224, 5, true)) {
List<RecognitionResponse> recognizedObjects =
ImageRecognition.toRecognitionResponse(recognitionService.recognizeTopK(inputImage));
List<RecognitionResponse> recognizedObjects = ImageRecognition
.toRecognitionResponse(recognitionService.recognizeTopK(inputImage));
// Draw the predicted labels on top of the input image.
byte[] augmentedImage = new ImageRecognitionAugmenter().apply(inputImage, recognizedObjects);
IOUtils.write(augmentedImage, new FileOutputStream("./image-recognition/target/image-augmented-inception.jpg"));
IOUtils.write(augmentedImage,
new FileOutputStream("./image-recognition/target/image-augmented-inception.jpg"));
String jsonRecognizedObjects = new JsonMapperFunction().apply(recognizedObjects);
System.out.println("inception result:" + jsonRecognizedObjects);
}
}
}

22
function/spring-image-recognition-function/src/test/java/org/springframework/cloud/fn/image/recognition/ImageRecognitionExample2.java
View File

@@ -45,8 +45,11 @@ public final class ImageRecognitionExample2 {
System.out.println(inceptions.recognizeTopK(inputImage));
System.out.println(ImageRecognition.toRecognitionResponse(inceptions.recognizeTopK(inputImage)));
IOUtils.write(augmenter.apply(inputImage, ImageRecognition.toRecognitionResponse(inceptions.recognizeTopK(inputImage))),
new FileOutputStream("./functions/function/image-recognition-function/target/image-augmented-inceptions.jpg"));
IOUtils.write(
augmenter.apply(inputImage,
ImageRecognition.toRecognitionResponse(inceptions.recognizeTopK(inputImage))),
new FileOutputStream(
"./functions/function/image-recognition-function/target/image-augmented-inceptions.jpg"));
inceptions.close();
ImageRecognition mobileNetV2 = ImageRecognition.mobileNetV2(
@@ -54,8 +57,11 @@ public final class ImageRecognitionExample2 {
224, 10, true);
System.out.println(mobileNetV2.recognizeMax(inputImage));
System.out.println(mobileNetV2.recognizeTopK(inputImage));
IOUtils.write(augmenter.apply(inputImage, ImageRecognition.toRecognitionResponse(mobileNetV2.recognizeTopK(inputImage))),
new FileOutputStream("./functions/function/image-recognition-function/target/image-augmented-mobilnetV2.jpg"));
IOUtils.write(
augmenter.apply(inputImage,
ImageRecognition.toRecognitionResponse(mobileNetV2.recognizeTopK(inputImage))),
new FileOutputStream(
"./functions/function/image-recognition-function/target/image-augmented-mobilnetV2.jpg"));
mobileNetV2.close();
ImageRecognition mobileNetV1 = ImageRecognition.mobileNetV1(
@@ -63,8 +69,12 @@ public final class ImageRecognitionExample2 {
224, 10, true);
System.out.println(mobileNetV1.recognizeMax(inputImage));
System.out.println(mobileNetV1.recognizeTopK(inputImage));
IOUtils.write(augmenter.apply(inputImage, ImageRecognition.toRecognitionResponse(mobileNetV1.recognizeTopK(inputImage))),
new FileOutputStream("./functions/function/image-recognition-function/target/image-augmented-mobilnetV1.jpg"));
IOUtils.write(
augmenter.apply(inputImage,
ImageRecognition.toRecognitionResponse(mobileNetV1.recognizeTopK(inputImage))),
new FileOutputStream(
"./functions/function/image-recognition-function/target/image-augmented-mobilnetV1.jpg"));
mobileNetV1.close();
}
}

22
function/spring-image-recognition-function/src/test/java/org/springframework/cloud/fn/image/recognition/SavedModelTest.java
View File

@@ -18,7 +18,6 @@ package org.springframework.cloud.fn.image.recognition;
import java.util.Map;
import com.google.protobuf.InvalidProtocolBufferException;
import org.tensorflow.SavedModelBundle;
import org.tensorflow.framework.MetaGraphDef;
@@ -40,11 +39,13 @@ public final class SavedModelTest {
*
*/
public static void main(String[] args) throws InvalidProtocolBufferException {
SavedModelBundle savedModelBundle =
SavedModelBundle.load("/Users/ctzolov/Downloads/ssd_mobilenet_v1_coco_2017_11_17/saved_model", "serve");
//SavedModelBundle.load("/Users/ctzolov/Downloads/aiy_vision_classifier_plants_V1_1/", "serve");
//SavedModelBundle savedModelBundle =
// SavedModelBundle.load("/Users/ctzolov/Downloads/mnasnet-a1/saved_model", "serve");
SavedModelBundle savedModelBundle = SavedModelBundle
.load("/Users/ctzolov/Downloads/ssd_mobilenet_v1_coco_2017_11_17/saved_model", "serve");
// SavedModelBundle.load("/Users/ctzolov/Downloads/aiy_vision_classifier_plants_V1_1/",
// "serve");
// SavedModelBundle savedModelBundle =
// SavedModelBundle.load("/Users/ctzolov/Downloads/mnasnet-a1/saved_model",
// "serve");
MetaGraphDef meta = MetaGraphDef.parseFrom(savedModelBundle.metaGraphDef());
@@ -54,10 +55,11 @@ public final class SavedModelTest {
savedModelBundle.session();
//Iterator<Operation> itr = savedModelBundle.graph().operations();
// Iterator<Operation> itr = savedModelBundle.graph().operations();
//
//while (itr.hasNext()) {
// System.out.println("Operation: " + itr.next());
//}
// while (itr.hasNext()) {
// System.out.println("Operation: " + itr.next());
// }
}
}