[api] Adds base64 image support for ImageTranslator

api/src/main/java/ai/djl/modality/cv/ImageFactory.java

@@ -17,12 +17,16 @@
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
+import java.io.ByteArrayInputStream;
 import java.io.IOException;
 import java.io.InputStream;
 import java.net.URI;
 import java.net.URL;
 import java.nio.file.Path;
 import java.nio.file.Paths;
+import java.util.Base64;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
 
 /**
  * {@code ImageFactory} contains image creation mechanism on top of different platforms like PC and
@@ -38,6 +42,8 @@ public abstract class ImageFactory {
         "ai.djl.android.core.BitmapImageFactory"
     };
 
+    private static final Pattern URL_PATTERN = Pattern.compile("^data:image/\\w+;base64,(.+)");
+
     private static ImageFactory factory = newInstance();
 
     private static ImageFactory newInstance() {
@@ -98,13 +104,22 @@ public Image fromUrl(URL url) throws IOException {
     }
 
     /**
-     * Gets {@link Image} from URL.
+     * Gets {@link Image} from string representation.
      *
-     * @param url the String represent URL to load from
+     * @param url the String represent URL or base64 encoded image to load from
      * @return {@link Image}
      * @throws IOException URL is not valid.
      */
     public Image fromUrl(String url) throws IOException {
+        Matcher m = URL_PATTERN.matcher(url);
+        if (m.matches()) {
+            // url="data:image/png;base64,..."
+            byte[] buf = Base64.getDecoder().decode(m.group(1));
+            try (InputStream is = new ByteArrayInputStream(buf)) {
+                return fromInputStream(is);
+            }
+        }
+
         URI uri = URI.create(url);
         if (uri.isAbsolute()) {
             return fromUrl(uri.toURL());

api/src/main/java/ai/djl/modality/cv/translator/BaseImageTranslatorFactory.java

@@ -15,8 +15,9 @@
 import ai.djl.modality.Input;
 import ai.djl.modality.Output;
 import ai.djl.modality.cv.Image;
-import ai.djl.modality.cv.translator.wrapper.FileImagePreProcesor;
+import ai.djl.modality.cv.translator.wrapper.FileImagePreProcessor;
 import ai.djl.modality.cv.translator.wrapper.InputStreamImagePreProcessor;
+import ai.djl.modality.cv.translator.wrapper.StringImagePreProcessor;
 import ai.djl.modality.cv.translator.wrapper.UrlImagePreProcessor;
 import ai.djl.translate.ExpansionTranslatorFactory;
 import ai.djl.translate.PreProcessor;
@@ -44,8 +45,9 @@ public abstract class BaseImageTranslatorFactory<O> extends ExpansionTranslatorF
             getPreprocessorExpansions() {
         Map<Type, Function<PreProcessor<Image>, PreProcessor<?>>> expansions =
                 new ConcurrentHashMap<>();
-        expansions.put(Path.class, FileImagePreProcesor::new);
+        expansions.put(Path.class, FileImagePreProcessor::new);
         expansions.put(URL.class, UrlImagePreProcessor::new);
+        expansions.put(String.class, StringImagePreProcessor::new);
         expansions.put(InputStream.class, InputStreamImagePreProcessor::new);
         return expansions;
     }

api/src/main/java/ai/djl/modality/cv/translator/ImageServingTranslator.java

@@ -23,6 +23,12 @@
 import ai.djl.translate.Translator;
 import ai.djl.translate.TranslatorContext;
 import ai.djl.util.JsonSerializable;
+import ai.djl.util.JsonUtils;
+
+import com.google.gson.JsonElement;
+import com.google.gson.JsonObject;
+import com.google.gson.JsonParseException;
+import com.google.gson.JsonPrimitive;
 
 import java.io.ByteArrayInputStream;
 import java.io.ByteArrayOutputStream;
@@ -80,7 +86,34 @@ public NDList processInput(TranslatorContext ctx, Input input) throws Exception
             if (data == null) {
                 throw new TranslateException("Input data is empty.");
             }
-            Image image = factory.fromInputStream(new ByteArrayInputStream(data.getAsBytes()));
+            String contentType = input.getProperty("Content-Type", null);
+            if (contentType != null) {
+                int pos = contentType.indexOf(';');
+                if (pos > 0) {
+                    contentType = contentType.substring(0, pos);
+                }
+            }
+            Image image;
+            if ("application/json".equalsIgnoreCase(contentType)) {
+                try {
+                    JsonElement element =
+                            JsonUtils.GSON.fromJson(data.getAsString(), JsonElement.class);
+                    if (element == null || !element.isJsonObject()) {
+                        throw new TranslateException("Invalid JsonObject input.");
+                    }
+                    JsonObject obj = element.getAsJsonObject();
+                    JsonPrimitive url = obj.getAsJsonPrimitive("image_url");
+                    if (url == null) {
+                        throw new TranslateException("Missing \"image_url\" in json.");
+                    }
+
+                    image = factory.fromUrl(url.getAsString());
+                } catch (JsonParseException e) {
+                    throw new TranslateException("Input is not a valid json.", e);
+                }
+            } else {
+                image = factory.fromInputStream(new ByteArrayInputStream(data.getAsBytes()));
+            }
             return translator.processInput(ctx, image);
         } catch (IOException e) {
             throw new TranslateException("Input is not an Image data type", e);

api/src/main/java/ai/djl/modality/cv/translator/wrapper/FileImagePreProcesor.java → api/src/main/java/ai/djl/modality/cv/translator/wrapper/FileImagePreProcessor.java

@@ -21,16 +21,16 @@
 import java.nio.file.Path;
 
 /** Built-in {@code PreProcessor} that provides image pre-processing from file path. */
-public class FileImagePreProcesor implements PreProcessor<Path> {
+public class FileImagePreProcessor implements PreProcessor<Path> {
 
     private PreProcessor<Image> preProcessor;
 
     /**
-     * Creates a {@code FileImagePreProcesor} instance.
+     * Creates a {@code FileImagePreProcessor} instance.
      *
-     * @param preProcessor a {@code {@link PreProcessor}} that can process image
+     * @param preProcessor a {@code PreProcessor} that can process image
      */
-    public FileImagePreProcesor(PreProcessor<Image> preProcessor) {
+    public FileImagePreProcessor(PreProcessor<Image> preProcessor) {
         this.preProcessor = preProcessor;
     }
 

api/src/main/java/ai/djl/modality/cv/translator/wrapper/StringImagePreProcessor.java

@@ -0,0 +1,44 @@
+/*
+ * Copyright 2024 Amazon.com, Inc. or its affiliates. All Rights Reserved.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License"). You may not use this file except in compliance
+ * with the License. A copy of the License is located at
+ *
+ * http://aws.amazon.com/apache2.0/
+ *
+ * or in the "license" file accompanying this file. This file is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES
+ * OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions
+ * and limitations under the License.
+ */
+package ai.djl.modality.cv.translator.wrapper;
+
+import ai.djl.modality.cv.Image;
+import ai.djl.modality.cv.ImageFactory;
+import ai.djl.ndarray.NDList;
+import ai.djl.translate.PreProcessor;
+import ai.djl.translate.TranslatorContext;
+
+/**
+ * Built-in {@code PreProcessor} that provides image pre-processing from url or base64 encoded
+ * string.
+ */
+public class StringImagePreProcessor implements PreProcessor<String> {
+
+    private PreProcessor<Image> preProcessor;
+
+    /**
+     * Creates a {@code StringImagePreProcessor} instance.
+     *
+     * @param preProcessor a {@code PreProcessor} that can process image
+     */
+    public StringImagePreProcessor(PreProcessor<Image> preProcessor) {
+        this.preProcessor = preProcessor;
+    }
+
+    /** {@inheritDoc} */
+    @Override
+    public NDList processInput(TranslatorContext ctx, String input) throws Exception {
+        Image image = ImageFactory.getInstance().fromUrl(input);
+        return preProcessor.processInput(ctx, image);
+    }
+}

api/src/test/java/ai/djl/modality/cv/translator/ImageClassificationTranslatorFactoryTest.java

@@ -41,7 +41,7 @@ public void setUp() {
 
     @Test
     public void testGetSupportedTypes() {
-        Assert.assertEquals(factory.getSupportedTypes().size(), 5);
+        Assert.assertEquals(factory.getSupportedTypes().size(), 6);
     }
 
     @Test
@@ -68,6 +68,10 @@ public void testNewInstance() {
                     factory.newInstance(Input.class, Output.class, model, arguments);
             Assert.assertTrue(translator5 instanceof ImageServingTranslator);
 
+            Translator<String, Classifications> translator6 =
+                    factory.newInstance(String.class, Classifications.class, model, arguments);
+            Assert.assertTrue(translator6 instanceof BasicTranslator);
+
             Assert.assertThrows(
                     IllegalArgumentException.class,
                     () -> factory.newInstance(Image.class, Output.class, model, arguments));

api/src/test/java/ai/djl/modality/cv/translator/InstanceSegmentationTranslatorFactoryTest.java

@@ -41,7 +41,7 @@ public void setUp() {
 
     @Test
     public void testGetSupportedTypes() {
-        Assert.assertEquals(factory.getSupportedTypes().size(), 5);
+        Assert.assertEquals(factory.getSupportedTypes().size(), 6);
     }
 
     @Test

api/src/test/java/ai/djl/modality/cv/translator/YoloTranslatorFactoryTest.java

@@ -41,7 +41,7 @@ public void setUp() {
 
     @Test
     public void testGetSupportedTypes() {
-        Assert.assertEquals(factory.getSupportedTypes().size(), 5);
+        Assert.assertEquals(factory.getSupportedTypes().size(), 6);
     }
 
     @Test

api/src/test/java/ai/djl/modality/cv/translator/YoloV5TranslatorFactoryTest.java

@@ -41,7 +41,7 @@ public void setUp() {
 
     @Test
     public void testGetSupportedTypes() {
-        Assert.assertEquals(factory.getSupportedTypes().size(), 5);
+        Assert.assertEquals(factory.getSupportedTypes().size(), 6);
     }
 
     @Test

api/src/test/java/ai/djl/modality/cv/translator/YoloV8TranslatorFactoryTest.java

@@ -41,7 +41,7 @@ public void setUp() {
 
     @Test
     public void testGetSupportedTypes() {
-        Assert.assertEquals(factory.getSupportedTypes().size(), 5);
+        Assert.assertEquals(factory.getSupportedTypes().size(), 6);
     }
 
     @Test

examples/src/main/java/ai/djl/examples/inference/cv/ActionRecognition.java

@@ -15,8 +15,6 @@
 import ai.djl.ModelException;
 import ai.djl.inference.Predictor;
 import ai.djl.modality.Classifications;
-import ai.djl.modality.cv.Image;
-import ai.djl.modality.cv.ImageFactory;
 import ai.djl.repository.zoo.Criteria;
 import ai.djl.repository.zoo.ZooModel;
 import ai.djl.training.util.ProgressBar;
@@ -26,6 +24,7 @@
 import org.slf4j.LoggerFactory;
 
 import java.io.IOException;
+import java.net.URL;
 
 /**
  * An example of inference using an action recognition model.
@@ -46,22 +45,20 @@ public static void main(String[] args) throws IOException, ModelException, Trans
     }
 
     public static Classifications predict() throws IOException, ModelException, TranslateException {
-        String url = "https://resources.djl.ai/images/action_dance.jpg";
-        Image img = ImageFactory.getInstance().fromUrl(url);
-
+        URL url = new URL("https://resources.djl.ai/images/action_dance.jpg");
         // Use DJL PyTorch model zoo model
-        Criteria<Image, Classifications> criteria =
+        Criteria<URL, Classifications> criteria =
                 Criteria.builder()
-                        .setTypes(Image.class, Classifications.class)
+                        .setTypes(URL.class, Classifications.class)
                         .optModelUrls(
                                 "djl://ai.djl.pytorch/Human-Action-Recognition-VIT-Base-patch16-224")
                         .optEngine("PyTorch")
                         .optProgress(new ProgressBar())
                         .build();
 
-        try (ZooModel<Image, Classifications> inception = criteria.loadModel();
-                Predictor<Image, Classifications> action = inception.newPredictor()) {
-            return action.predict(img);
+        try (ZooModel<URL, Classifications> inception = criteria.loadModel();
+                Predictor<URL, Classifications> action = inception.newPredictor()) {
+            return action.predict(url);
         }
     }
 }

examples/src/main/java/ai/djl/examples/inference/cv/MaskDetection.java

@@ -56,9 +56,9 @@ public static DetectedObjects predict() throws IOException, ModelException, Tran
 
         // modelUrl can be replaced to local onnx model file
         String modelUrl = "https://resources.djl.ai/demo/onnxruntime/face_mask_detection.zip";
-        Criteria<Image, DetectedObjects> criteria =
+        Criteria<String, DetectedObjects> criteria =
                 Criteria.builder()
-                        .setTypes(Image.class, DetectedObjects.class)
+                        .setTypes(String.class, DetectedObjects.class)
                         .optModelUrls(modelUrl)
                         .optEngine("OnnxRuntime")
                         .optTranslatorFactory(new YoloV5TranslatorFactory())
@@ -67,9 +67,9 @@ public static DetectedObjects predict() throws IOException, ModelException, Tran
                         .optArgument("rescale", true) // post process
                         .build();
 
-        try (ZooModel<Image, DetectedObjects> model = criteria.loadModel()) {
-            try (Predictor<Image, DetectedObjects> predictor = model.newPredictor()) {
-                DetectedObjects detection = predictor.predict(img);
+        try (ZooModel<String, DetectedObjects> model = criteria.loadModel()) {
+            try (Predictor<String, DetectedObjects> predictor = model.newPredictor()) {
+                DetectedObjects detection = predictor.predict(imageUrl);
                 String outputDir = "build/output";
                 saveBoundingBoxImage(img, detection, outputDir);
                 return detection;

examples/src/main/java/ai/djl/examples/inference/cv/Yolov8Detection.java

@@ -50,9 +50,9 @@ public static DetectedObjects predict() throws IOException, ModelException, Tran
 
         // Use DJL OnnxRuntime model zoo model, model can be found:
         // https://mlrepo.djl.ai/model/cv/object_detection/ai/djl/onnxruntime/yolov8n/0.0.1/yolov8n.zip
-        Criteria<Image, DetectedObjects> criteria =
+        Criteria<Path, DetectedObjects> criteria =
                 Criteria.builder()
-                        .setTypes(Image.class, DetectedObjects.class)
+                        .setTypes(Path.class, DetectedObjects.class)
                         .optModelUrls("djl://ai.djl.onnxruntime/yolov8n")
                         .optEngine("OnnxRuntime")
                         .optArgument("width", 640)
@@ -68,12 +68,12 @@ public static DetectedObjects predict() throws IOException, ModelException, Tran
                         .optProgress(new ProgressBar())
                         .build();
 
-        try (ZooModel<Image, DetectedObjects> model = criteria.loadModel();
-                Predictor<Image, DetectedObjects> predictor = model.newPredictor()) {
+        try (ZooModel<Path, DetectedObjects> model = criteria.loadModel();
+                Predictor<Path, DetectedObjects> predictor = model.newPredictor()) {
             Path outputPath = Paths.get("build/output");
             Files.createDirectories(outputPath);
 
-            DetectedObjects detection = predictor.predict(img);
+            DetectedObjects detection = predictor.predict(imgPath);
             if (detection.getNumberOfObjects() > 0) {
                 img.drawBoundingBoxes(detection);
                 Path output = outputPath.resolve("yolov8_detected.png");

integration/src/main/java/ai/djl/integration/tests/modality/cv/BufferedImageFactoryTest.java

@@ -24,19 +24,26 @@
 import org.testng.Assert;
 import org.testng.annotations.Test;
 
+import java.io.ByteArrayOutputStream;
 import java.io.IOException;
+import java.util.Base64;
 
 public class BufferedImageFactoryTest {
 
     @Test
     public void testLoadImage() throws IOException {
         try (NDManager manager = NDManager.newBaseManager(TestUtils.getEngine())) {
             ImageFactory factory = ImageFactory.getInstance();
-            Image img =
-                    factory.fromUrl(
-                            "https://github.com/deepjavalibrary/djl/raw/master/examples/src/test/resources/dog_bike_car.jpg");
+            Image img = factory.fromUrl("https://resources.djl.ai/images/dog_bike_car.jpg");
             NDArray array = img.toNDArray(manager);
             Assert.assertEquals(new Shape(img.getHeight(), img.getWidth(), 3), array.getShape());
+
+            try (ByteArrayOutputStream bos = new ByteArrayOutputStream()) {
+                img.save(bos, "png");
+                String data = Base64.getEncoder().encodeToString(bos.toByteArray());
+                Image img2 = factory.fromUrl("data:image/png;base64," + data);
+                Assert.assertEquals(img2.getWidth(), 768);
+            }
         }
     }