RandomResizedCrop

docs/api/api_python/dataset_vision/mindspore.dataset.vision.RandomResizedCrop.rst

@@ -31,4 +31,4 @@ mindspore.dataset.vision.RandomResizedCrop
         - **ValueError** - 当 `scale` 为负数。
         - **ValueError** - 当 `ratio` 为负数。
         - **ValueError** - 当 `max_attempts` 不为正数。
-        - **RuntimeError** - 当输入图像的shape不为<H, W>或<H, W, C>。
+        - **RuntimeError** - 当输入图像的shape不为<H, W>或<..., H, W, C>。

mindspore/ccsrc/minddata/dataset/kernels/image/random_crop_and_resize_op.cc

@@ -1,5 +1,5 @@
 /**
- * Copyright 2021 Huawei Technologies Co., Ltd
+ * Copyright 2020-2022 Huawei Technologies Co., Ltd
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -14,9 +14,12 @@
  * limitations under the License.
  */
 #include "minddata/dataset/kernels/image/random_crop_and_resize_op.h"
+
 #include <limits>
 #include <random>
+#include <vector>
 
+#include "minddata/dataset/kernels/data/data_utils.h"
 #include "minddata/dataset/kernels/image/image_utils.h"
 #include "minddata/dataset/util/random.h"
 #include "minddata/dataset/util/status.h"
@@ -47,16 +50,20 @@ RandomCropAndResizeOp::RandomCropAndResizeOp(int32_t target_height, int32_t targ
 
 Status RandomCropAndResizeOp::Compute(const TensorRow &input, TensorRow *output) {
   IO_CHECK_VECTOR(input, output);
-  if (input.size() != 1) {
+
-    for (size_t i = 0; i < input.size() - 1; i++) {
+  for (size_t i = 0; i < input.size(); i++) {
-      if (input[i]->Rank() != 2 && input[i]->Rank() != 3) {
+    if (input[i]->Rank() < kMinImageRank) {
-        std::string err_msg = "RandomCropAndResizeOp: image shape is not <H,W,C> or <H, W>, but got rank:" +
+      RETURN_STATUS_UNEXPECTED("RandomResizedCrop: input tensor should have at least 2 dimensions, but got: " +
-                              std::to_string(input[i]->Rank());
+                               std::to_string(input[i]->Rank()));
-        RETURN_STATUS_UNEXPECTED(err_msg);
+    }
-      }
+    if (i < input.size() - 1) {
-      if (input[i]->shape()[0] != input[i + 1]->shape()[0] || input[i]->shape()[1] != input[i + 1]->shape()[1]) {
+      std::vector<dsize_t> size;
+      std::vector<dsize_t> next_size;
+      RETURN_IF_NOT_OK(ImageSize(input[i], &size));
+      RETURN_IF_NOT_OK(ImageSize(input[i + 1], &next_size));
+      if (size[0] != next_size[0] || size[1] != next_size[1]) {
         RETURN_STATUS_UNEXPECTED(
-          "RandomCropAndResizeOp: Input images in different column of each row must have the same size.");
+          "RandomCropAndResizeOp: Input tensor in different columns of each row must have the same size.");
       }
     }
   }
@@ -66,18 +73,50 @@ Status RandomCropAndResizeOp::Compute(const TensorRow &input, TensorRow *output)
   int crop_height = 0;
   int crop_width = 0;
   for (size_t i = 0; i < input.size(); i++) {
-    RETURN_IF_NOT_OK(ValidateImageRank("RandomCropAndResize", static_cast<int32_t>(input[i]->shape().Size())));
+    auto input_shape = input[i]->shape();
-    int h_in = static_cast<int>(input[i]->shape()[0]);
+    std::vector<dsize_t> size;
-    int w_in = static_cast<int>(input[i]->shape()[1]);
+    RETURN_IF_NOT_OK(ImageSize(input[i], &size));
+    int h_in = size[0];
+    int w_in = size[1];
     if (i == 0) {
       RETURN_IF_NOT_OK(GetCropBox(h_in, w_in, &x, &y, &crop_height, &crop_width));
     }
-    RETURN_IF_NOT_OK(CropAndResize(input[i], &(*output)[i], x, y, crop_height, crop_width, target_height_,
+    if (input[i]->Rank() <= kDefaultImageRank) {
-                                   target_width_, interpolation_));
+      RETURN_IF_NOT_OK(CropAndResize(input[i], &(*output)[i], x, y, crop_height, crop_width, target_height_,
+                                     target_width_, interpolation_));
+    } else if (input[i]->Rank() > kDefaultImageRank) {
+      dsize_t num_batch = input[i]->Size() / (input_shape[-3] * input_shape[-2] * input_shape[-1]);
+      TensorShape new_shape({num_batch, input_shape[-3], input_shape[-2], input_shape[-1]});
+      RETURN_IF_NOT_OK(input[i]->Reshape(new_shape));
+      // split [N, H, W, C] to N [H, W, C], and Resize N [H, W, C]
+      std::vector<std::shared_ptr<Tensor>> input_vector_hwc, output_vector_hwc;
+      RETURN_IF_NOT_OK(BatchTensorToTensorVector(input[i], &input_vector_hwc));
+      for (auto input_hwc : input_vector_hwc) {
+        std::shared_ptr<Tensor> output_img;
+        RETURN_IF_NOT_OK(CropAndResize(input_hwc, &output_img, x, y, crop_height, crop_width, target_height_,
+                                       target_width_, interpolation_));
+        output_vector_hwc.push_back(output_img);
+      }
+      RETURN_IF_NOT_OK(TensorVectorToBatchTensor(output_vector_hwc, &(*output)[i]));
+      auto output_shape = ComputeOutputShape(input_shape, target_height_, target_width_);
+      RETURN_IF_NOT_OK((*output)[i]->Reshape(output_shape));
+    }
   }
   return Status::OK();
 }
 
+TensorShape RandomCropAndResizeOp::ComputeOutputShape(const TensorShape &input, int32_t target_height,
+                                                      int32_t target_width) {
+  auto out_shape_vec = input.AsVector();
+  auto size = out_shape_vec.size();
+  int32_t kHeightIdx = -3;
+  int32_t kWidthIdx = -2;
+  out_shape_vec[size + kHeightIdx] = target_height_;
+  out_shape_vec[size + kWidthIdx] = target_width_;
+  TensorShape out = TensorShape(out_shape_vec);
+  return out;
+}
+
 Status RandomCropAndResizeOp::OutputShape(const std::vector<TensorShape> &inputs, std::vector<TensorShape> &outputs) {
   RETURN_IF_NOT_OK(TensorOp::OutputShape(inputs, outputs));
   outputs.clear();
@@ -87,12 +126,16 @@ Status RandomCropAndResizeOp::OutputShape(const std::vector<TensorShape> &inputs
   }
   if (inputs[0].Rank() == 3) {
     (void)outputs.emplace_back(out.AppendDim(inputs[0][2]));
+  } else if (inputs[0].Rank() > kDefaultImageRank) {
+    auto out_shape = ComputeOutputShape(inputs[0], target_height_, target_width_);
+    (void)outputs.emplace_back(out_shape);
   }
   if (!outputs.empty()) {
     return Status::OK();
   }
   return Status(StatusCode::kMDUnexpectedError,
-                "RandomCropAndResize: invalid input shape, expected 2D or 3D input, but got input dimension is: " +
+                "RandomCropAndResize: input tensor should have at least 2 dimensions, "
+                "but got: " +
                   std::to_string(inputs[0].Rank()));
 }
 Status RandomCropAndResizeOp::GetCropBox(int h_in, int w_in, int *x, int *y, int *crop_height, int *crop_width) {

mindspore/ccsrc/minddata/dataset/kernels/image/random_crop_and_resize_op.h

@@ -22,7 +22,6 @@
 #include <string>
 
 #include "minddata/dataset/core/tensor.h"
-#include "minddata/dataset/kernels/image/image_utils.h"
 #include "minddata/dataset/kernels/tensor_op.h"
 #include "minddata/dataset/util/status.h"
 
@@ -58,6 +57,8 @@ class RandomCropAndResizeOp : public TensorOp {
 
   Status OutputShape(const std::vector<TensorShape> &inputs, std::vector<TensorShape> &outputs) override;
 
+  TensorShape ComputeOutputShape(const TensorShape &input, int32_t target_height, int32_t target_width);
+
   Status GetCropBox(int h_in, int w_in, int *x, int *y, int *crop_height, int *crop_width);
 
   std::string Name() const override { return kRandomCropAndResizeOp; }

mindspore/python/mindspore/dataset/vision/__init__.py

@@ -77,7 +77,7 @@ from .transforms import AdjustBrightness, AdjustContrast, AdjustGamma, AdjustHue
     Affine, AutoAugment, AutoContrast, BoundingBoxAugment, CenterCrop, ConvertColor, Crop, CutMixBatch, CutOut, \
     Decode, Equalize, Erase, FiveCrop, GaussianBlur, Grayscale, HorizontalFlip, HsvToRgb, HWC2CHW, Invert, \
     LinearTransformation, MixUp, MixUpBatch, Normalize, NormalizePad, Pad, PadToSize, Perspective, Posterize, \
-    RandomAdjustSharpness, RandomAffine, RandomAutoContrast, RandomColor, RandomColorAdjust, RandomCrop, \
+    RandAugment, RandomAdjustSharpness, RandomAffine, RandomAutoContrast, RandomColor, RandomColorAdjust, RandomCrop, \
     RandomCropDecodeResize, RandomCropWithBBox, RandomEqualize, RandomErasing, RandomGrayscale, RandomHorizontalFlip, \
     RandomHorizontalFlipWithBBox, RandomInvert, RandomLighting, RandomPerspective, RandomPosterize, RandomResizedCrop, \
     RandomResizedCropWithBBox, RandomResize, RandomResizeWithBBox, RandomRotation, RandomSelectSubpolicy, \

mindspore/python/mindspore/dataset/vision/c_transforms.py

@@ -1749,7 +1749,7 @@ class RandomResizedCrop(ImageTensorOperation):
         ValueError: If `scale` is negative.
         ValueError: If `ratio` is negative.
         ValueError: If `max_attempts` is not positive.
-        RuntimeError: If given tensor shape is not <H, W> or <H, W, C>.
+        RuntimeError: If given tensor shape is not <H, W> or <..., H, W, C>.
 
     Supported Platforms:
         ``CPU``

mindspore/python/mindspore/dataset/vision/transforms.py

@@ -1830,7 +1830,7 @@ class RandAugment(ImageTensorOperation):
         TypeError: If `num_ops` is not of type int.
         TypeError: If `magnitude` is not of type int.
         TypeError: If `num_magnitude_bins` is not of type int.
-        TypeError: If `interpolation` not of type int.
+        TypeError: If `interpolation` not of type :class:`mindspore.dataset.vision.Inter`.
         TypeError: If `fill_value` is not an int or a tuple of length 3.
         RuntimeError: If given tensor shape is not <H, W, C>.
 
@@ -2951,7 +2951,7 @@ class RandomResizedCrop(ImageTensorOperation, PyTensorOperation):
         ValueError: If `scale` is negative.
         ValueError: If `ratio` is negative.
         ValueError: If `max_attempts` is not positive.
-        RuntimeError: If given tensor shape is not <H, W> or <H, W, C>.
+        RuntimeError: If given tensor shape is not <H, W> or <..., H, W, C>.
 
     Supported Platforms:
         ``CPU``

tests/ut/cpp/dataset/random_crop_and_resize_op_test.cc

@@ -13,9 +13,11 @@
  * See the License for the specific language governing permissions and
  * limitations under the License.
  */
+#include <random>
+
 #include "common/common.h"
 #include "common/cvop_common.h"
-#include <random>
+#include "minddata/dataset/kernels/data/data_utils.h"
 #include "minddata/dataset/kernels/image/random_crop_and_resize_op.h"
 #include "utils/log_adapter.h"
 
@@ -114,4 +116,4 @@ TEST_F(MindDataTestRandomCropAndResizeOp, TestOpSimpleTest3) {
   }
 
   MS_LOG(INFO) << "RandomCropAndResizeOp simple test finished";
-}
+}

tests/ut/python/dataset/test_rand_augment.py

@@ -18,8 +18,8 @@ Testing RandAugment in DE
 import numpy as np
 
 import mindspore.dataset as ds
-from mindspore.dataset.vision.transforms import Decode, RandAugment, Resize
+from mindspore.dataset.vision import Decode, RandAugment, Resize
-from mindspore.dataset.vision.utils import Inter
+from mindspore.dataset.vision import Inter
 from mindspore import log as logger
 from util import visualize_image, visualize_list, diff_mse
 

tests/ut/python/dataset/test_random_crop_and_resize.py

@@ -484,6 +484,64 @@ def test_random_crop_and_resize_07():
         num_iter += 1
 
 
+def test_random_crop_and_resize_08():
+    """
+    Feature: RandomCropAndResize
+    Description: Test RandomCropAndResize with 4 dim image
+    Expectation: The data is processed successfully
+    """
+    logger.info("test_random_crop_and_resize_08")
+
+    original_seed = config_get_set_seed(5)
+    original_worker = config_get_set_num_parallel_workers(1)
+
+    data = np.random.randint(0, 255, (3, 3, 4, 3), np.uint8)
+    res1 = [[[83, 24, 209], [114, 181, 190]], [[200, 201, 36], [154, 13, 117]]]
+    res2 = [[[158, 140, 182], [104, 154, 109]], [[230, 79, 193], [87, 170, 223]]]
+    res3 = [[[179, 202, 143], [150, 178, 67]], [[20, 94, 159], [253, 151, 82]]]
+    expected_result = np.array([res1, res2, res3], dtype=np.uint8)
+
+    random_crop_and_resize_op = vision.RandomResizedCrop((2, 2))
+    output = random_crop_and_resize_op(data)
+
+    mse = diff_mse(output, expected_result)
+    assert mse < 0.0001
+    assert output.shape[-2] == 2
+    assert output.shape[-3] == 2
+
+    ds.config.set_seed(original_seed)
+    ds.config.set_num_parallel_workers(original_worker)
+
+
+def test_random_crop_and_resize_pipeline():
+    """
+    Feature: RandomCropAndResize
+    Description: Test RandomCropAndResize with 4 dim image
+    Expectation: The data is processed successfully
+    """
+    logger.info("Test RandomCropAndResize pipeline with 4 dimension input")
+
+    original_seed = config_get_set_seed(5)
+    original_worker = config_get_set_num_parallel_workers(1)
+
+    data = np.random.randint(0, 255, (1, 3, 3, 4, 3), np.uint8)
+    res1 = [[[83, 24, 209], [114, 181, 190]], [[200, 201, 36], [154, 13, 117]]]
+    res2 = [[[158, 140, 182], [104, 154, 109]], [[230, 79, 193], [87, 170, 223]]]
+    res3 = [[[179, 202, 143], [150, 178, 67]], [[20, 94, 159], [253, 151, 82]]]
+    expected_result = np.array([[res1, res2, res3]], dtype=np.uint8)
+
+    random_crop_and_resize = vision.RandomResizedCrop((2, 2))
+    dataset = ds.NumpySlicesDataset(data, column_names=["image"], shuffle=False)
+    dataset = dataset.map(input_columns=["image"], operations=random_crop_and_resize)
+
+    for i, item in enumerate(dataset.create_dict_iterator(output_numpy=True)):
+        mse = diff_mse(item["image"], expected_result[i])
+        assert mse < 0.0001
+
+    ds.config.set_seed(original_seed)
+    ds.config.set_num_parallel_workers(original_worker)
+
+
 def test_random_crop_and_resize_eager_error_01():
     """
     Feature: RandomCropAndResize op
@@ -534,5 +592,7 @@ if __name__ == "__main__":
     test_random_crop_and_resize_06()
     test_random_crop_and_resize_comp(True)
     test_random_crop_and_resize_07()
+    test_random_crop_and_resize_08()
+    test_random_crop_and_resize_pipeline()
     test_random_crop_and_resize_eager_error_01()
     test_random_crop_and_resize_eager_error_02()