Add UINT8 datatype support to Java (#8401)

Add UINT8 datatype support
Add inference test for UINT8 model

java/src/main/java/ai/onnxruntime/OnnxJavaType.java

@@ -16,9 +16,10 @@ public enum OnnxJavaType {
   INT64(6, long.class, 8),
   BOOL(7, boolean.class, 1),
   STRING(8, String.class, 4),
+  UINT8(9, byte.class, 1),
   UNKNOWN(0, Object.class, 0);
 
-  private static final OnnxJavaType[] values = new OnnxJavaType[9];
+  private static final OnnxJavaType[] values = new OnnxJavaType[10];
 
   static {
     for (OnnxJavaType ot : OnnxJavaType.values()) {
@@ -62,6 +63,7 @@ public enum OnnxJavaType {
   public static OnnxJavaType mapFromOnnxTensorType(OnnxTensorType onnxValue) {
     switch (onnxValue) {
       case ONNX_TENSOR_ELEMENT_DATA_TYPE_UINT8:
+        return OnnxJavaType.UINT8;
       case ONNX_TENSOR_ELEMENT_DATA_TYPE_INT8:
         return OnnxJavaType.INT8;
       case ONNX_TENSOR_ELEMENT_DATA_TYPE_UINT16:

java/src/main/java/ai/onnxruntime/OnnxMap.java

@@ -77,6 +77,7 @@ public class OnnxMap implements OnnxValue {
           return OnnxMapValueType.LONG;
         case STRING:
           return OnnxMapValueType.STRING;
+        case UINT8:
         case INT8:
         case INT16:
         case INT32:

java/src/main/java/ai/onnxruntime/OnnxSequence.java

@@ -100,6 +100,7 @@ public class OnnxSequence implements OnnxValue {
           list.addAll(Arrays.asList(strings));
           return list;
         case BOOL:
+        case UINT8:
         case INT8:
         case INT16:
         case INT32:

java/src/main/java/ai/onnxruntime/OnnxTensor.java

@@ -78,6 +78,7 @@ public class OnnxTensor implements OnnxValue {
           return getFloat(OnnxRuntime.ortApiHandle, nativeHandle, info.onnxType.value);
         case DOUBLE:
           return getDouble(OnnxRuntime.ortApiHandle, nativeHandle);
+        case UINT8:
         case INT8:
           return getByte(OnnxRuntime.ortApiHandle, nativeHandle, info.onnxType.value);
         case INT16:
@@ -744,6 +745,7 @@ public class OnnxTensor implements OnnxValue {
         case DOUBLE:
           tmp = buffer.asDoubleBuffer().put((DoubleBuffer) data);
           break;
+        case UINT8:
         case INT8:
           // buffer is already a ByteBuffer, no cast needed.
           tmp = buffer.put((ByteBuffer) data);

java/src/main/java/ai/onnxruntime/OrtUtil.java

@@ -434,6 +434,7 @@ public final class OrtUtil {
         double[] doubleArr = new double[1];
         doubleArr[0] = (Double) data;
         return doubleArr;
+      case UINT8:
       case INT8:
         byte[] byteArr = new byte[1];
         byteArr[0] = (Byte) data;

java/src/main/java/ai/onnxruntime/TensorInfo.java

@@ -80,6 +80,8 @@ public class TensorInfo implements ValueInfo {
           return OnnxTensorType.ONNX_TENSOR_ELEMENT_DATA_TYPE_DOUBLE;
         case INT8:
           return OnnxTensorType.ONNX_TENSOR_ELEMENT_DATA_TYPE_INT8;
+        case UINT8:
+          return OnnxTensorType.ONNX_TENSOR_ELEMENT_DATA_TYPE_UINT8;
         case INT16:
           return OnnxTensorType.ONNX_TENSOR_ELEMENT_DATA_TYPE_INT16;
         case INT32:
@@ -179,6 +181,7 @@ public class TensorInfo implements ValueInfo {
         return OrtUtil.newFloatArray(shape);
       case DOUBLE:
         return OrtUtil.newDoubleArray(shape);
+      case UINT8:
       case INT8:
         return OrtUtil.newByteArray(shape);
       case INT16:

java/src/test/java/ai/onnxruntime/InferenceTest.java

@@ -1247,6 +1247,28 @@ public class InferenceTest {
     }
   }
 
+  @Test
+  public void testModelInputUINT8() throws OrtException {
+    String modelPath = getResourcePath("/test_types_UINT8.pb").toString();
+
+    try (OrtEnvironment env = OrtEnvironment.getEnvironment("testModelInputUINT8");
+        SessionOptions options = new SessionOptions();
+        OrtSession session = env.createSession(modelPath, options)) {
+      String inputName = session.getInputNames().iterator().next();
+      Map<String, OnnxTensor> container = new HashMap<>();
+      byte[] flatInput = new byte[] {1, 2, -3, Byte.MIN_VALUE, Byte.MAX_VALUE};
+      ByteBuffer data = ByteBuffer.wrap(flatInput);
+      long[] shape = new long[] {1, 5};
+      OnnxTensor ov = OnnxTensor.createTensor(env, data, shape, OnnxJavaType.UINT8);
+      container.put(inputName, ov);
+      try (OrtSession.Result res = session.run(container)) {
+        byte[] resultArray = TestHelpers.flattenByte(res.get(0).getValue());
+        assertArrayEquals(flatInput, resultArray);
+      }
+      OnnxValue.close(container);
+    }
+  }
+
   @Test
   public void testModelInputINT16() throws OrtException {
     // model takes 1x5 input of fixed type, echoes back

java/src/test/java/ai/onnxruntime/TensorCreationTest.java

@@ -4,6 +4,7 @@
  */
 package ai.onnxruntime;
 
+import java.nio.ByteBuffer;
 import org.junit.jupiter.api.Assertions;
 import org.junit.jupiter.api.Test;
 
@@ -112,4 +113,16 @@ public class TensorCreationTest {
       }
     }
   }
+
+  @Test
+  public void testUint8Creation() throws OrtException {
+    try (OrtEnvironment env = OrtEnvironment.getEnvironment()) {
+      byte[] buf = new byte[] {0, 1};
+      ByteBuffer data = ByteBuffer.wrap(buf);
+      long[] shape = new long[] {2};
+      try (OnnxTensor t = OnnxTensor.createTensor(env, data, shape, OnnxJavaType.UINT8)) {
+        Assertions.assertArrayEquals(buf, (byte[]) t.getValue());
+      }
+    }
+  }
 }