1 /*
2 * Copyright (C) 2018 The Android Open Source Project
3 *
4 * Licensed under the Apache License, Version 2.0 (the "License");
5 * you may not use this file except in compliance with the License.
6 * You may obtain a copy of the License at
7 *
8 * http://www.apache.org/licenses/LICENSE-2.0
9 *
10 * Unless required by applicable law or agreed to in writing, software
11 * distributed under the License is distributed on an "AS IS" BASIS,
12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 * See the License for the specific language governing permissions and
14 * limitations under the License.
15 */
16
17 // Contains the implementation of the operations.
18
19 #define LOG_TAG "Operations"
20
21 #include "Operations.h"
22 #include "CpuOperationUtils.h"
23
24 #include "tensorflow/contrib/lite/kernels/internal/reference/reference_ops.h"
25
26 namespace android {
27 namespace nn {
28
stridedSliceGeneric(const uint8_t * inputData,const Shape & inputShape,const int32_t * beginData,const int32_t * endData,const int32_t * stridesData,int32_t beginMask,int32_t endMask,int32_t shrinkAxisMask,uint8_t * outputData,const Shape & outputShape)29 bool stridedSliceGeneric(const uint8_t* inputData, const Shape& inputShape,
30 const int32_t* beginData, const int32_t* endData,
31 const int32_t* stridesData,
32 int32_t beginMask, int32_t endMask, int32_t shrinkAxisMask,
33 uint8_t* outputData, const Shape& outputShape) {
34 // This Op only supports 1-4D cases and since we use the reference 4D
35 // implementation, the 1-3D tensors are mapped to 4D.
36 const int kMaxDim = 4;
37
38 std::vector<int> starts;
39 std::vector<int> stops;
40 std::vector<int> strides;
41
42 int32_t numInputDims = static_cast<int32_t>(getNumberOfDimensions(inputShape));
43 for (int32_t idx = numInputDims - 1; idx >= 0; --idx) {
44 int32_t dim = static_cast<int32_t>(getSizeOfDimension(inputShape, idx));
45 int32_t stride = stridesData[idx];
46 // stride value has to be non-zero
47 NN_OPS_CHECK(stride != 0);
48 bool positiveStride = stride > 0;
49
50 int32_t begin = beginMask & (1 << idx)
51 ? positiveStride ? 0 : dim - 1
52 : ClampedIndex(beginData[idx], dim, positiveStride);
53 int32_t end = endMask & (1 << idx)
54 ? positiveStride ? dim : -1
55 : ClampedIndex(endData[idx], dim, positiveStride);
56
57 starts.emplace_back(begin);
58 stops.emplace_back(end);
59 strides.emplace_back(stride);
60 }
61
62 for (int i = numInputDims; i < kMaxDim; i++) {
63 starts.emplace_back(0);
64 stops.emplace_back(1);
65 strides.emplace_back(1);
66 }
67
68 beginMask = ReverseMaskBits(beginMask, numInputDims);
69 endMask = ReverseMaskBits(endMask, numInputDims);
70 shrinkAxisMask = ReverseMaskBits(shrinkAxisMask, numInputDims);
71
72 if (inputShape.type == OperandType::TENSOR_FLOAT32) {
73 tflite::reference_ops::StridedSlice(
74 reinterpret_cast<const float*>(inputData),
75 convertShapeToDims(inputShape),
76 beginMask, endMask, shrinkAxisMask,
77 starts, stops, strides,
78 reinterpret_cast<float*>(outputData),
79 convertShapeToDims(outputShape));
80 } else if (inputShape.type == OperandType::TENSOR_QUANT8_ASYMM) {
81 tflite::reference_ops::StridedSlice(
82 reinterpret_cast<const uint8_t*>(inputData),
83 convertShapeToDims(inputShape),
84 beginMask, endMask, shrinkAxisMask,
85 starts, stops, strides,
86 reinterpret_cast<uint8_t*>(outputData),
87 convertShapeToDims(outputShape));
88 } else {
89 LOG(ERROR) << "Unsupported data type";
90 return false;
91 }
92
93 return true;
94 }
95
96 } // namespace nn
97 } // namespace android
98