diff options
author | Georgios Pinitas <georgios.pinitas@arm.com> | 2018-07-16 15:41:27 +0100 |
---|---|---|
committer | Anthony Barbier <anthony.barbier@arm.com> | 2018-11-02 16:54:54 +0000 |
commit | ae54e026c86aec7d6819ee3ef76372c1a3c92467 (patch) | |
tree | 83dd35aad5c524f99819d144583d5566923f1a97 /arm_compute/runtime/NEON/functions/NEDepthConcatenateLayer.h | |
parent | b0b37177f190a261b338cca53b6c6136eea14ba1 (diff) | |
download | ComputeLibrary-ae54e026c86aec7d6819ee3ef76372c1a3c92467.tar.gz |
COMPMID-1364: Add support for NHWC in NEDepthConcatenateLayer
Change-Id: I4f8e46d1c79afa9284f2c6dc00383c453a8e7bd5
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/140165
Reviewed-by: Giorgio Arena <giorgio.arena@arm.com>
Reviewed-by: Pablo Tello <pablo.tello@arm.com>
Tested-by: Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'arm_compute/runtime/NEON/functions/NEDepthConcatenateLayer.h')
-rw-r--r-- | arm_compute/runtime/NEON/functions/NEDepthConcatenateLayer.h | 30 |
1 files changed, 27 insertions, 3 deletions
diff --git a/arm_compute/runtime/NEON/functions/NEDepthConcatenateLayer.h b/arm_compute/runtime/NEON/functions/NEDepthConcatenateLayer.h index eefb5fa362..e2162ef042 100644 --- a/arm_compute/runtime/NEON/functions/NEDepthConcatenateLayer.h +++ b/arm_compute/runtime/NEON/functions/NEDepthConcatenateLayer.h @@ -49,10 +49,34 @@ public: NEDepthConcatenateLayer(); /** Initialise the kernel's inputs vector and output. * - * @param[in,out] inputs_vector The vectors containing all the tensors to concatenate. Data types supported: F16/F32. - * @param[out] output Output tensor. Data types supported: Same as @p inputs_vector. + * @param[in,out] inputs_vector The vectors containing all the tensors to concatenate. Data types supported: QASYMM8/F16/F32. + * Input dimensions might differ for each input for the first three dimensions (width, height, depth) + * and must match for the rest. + * Note that the difference between the minimum and maximum width and height among the input tensors + * must be divisible by 2 otherwise it is not clear how padding should be added on the inputs' width and + * height when they are less than the maximum input sizes. + * @param[out] output Output tensor. Data types supported: Same as @p input. + * Output tensor dimensions match the inputs' ones from the fourth dimension and above, + * while width and height are the maximum width and height of the input tensors. + * Finally, depth is the sum of the input depths. */ - void configure(std::vector<ITensor *> inputs_vector, ITensor *output); + void configure(const std::vector<ITensor *> &inputs_vector, ITensor *output); + /** Static function to check if given info will lead to a valid configuration of @ref NEDepthConcatenateLayer + * + * @param[in] inputs_vector The vectors containing all the tensors to concatenate. Data types supported: QASYMM8/F16/F32. + * Input dimensions might differ for each input for the first three dimensions (width, height, depth) + * and must match for the rest. + * Note that the difference between the minimum and maximum width and height among the input tensors + * must be divisible by 2 otherwise it is not clear how padding should be added on the inputs' width and + * height when they are less than the maximum input sizes. + * @param[in] output Output tensor. Data types supported: Same as @p input. + * Output tensor dimensions match the inputs' ones from the fourth dimension and above, + * while width and height are the maximum width and height of the input tensors. + * Finally, depth is the sum of the input depths. + * + * @return a status + */ + static Status validate(const std::vector<ITensorInfo *> &inputs_vector, const ITensorInfo *output); // Inherited methods overridden: void run() override; |