Merge branch 'master' of https://github.com/NVIDIA/CUDALibrarySamples

Author: springer13

nvJPEG/README.md

@@ -14,6 +14,8 @@ Image Encoding and Decoding from NVJPEG Library
 
 [JPEG Image Decoder MultipleInstances](nvJPEG-Decoder-MultipleInstances/)
 
+[JPEG Image Decoder Backend and ROI](nvJPEG-Decoder-Backend-ROI/)
+
 [Image Resize](Image-Resize/)
 
 [Image Resize Watermarking](Image-Resize-WaterMark/)
@@ -36,5 +38,5 @@ x86_64
 
 # Prerequisites
 - A Linux system with recent NVIDIA drivers.
-- Install the [CUDA 11.0 toolkit](https://developer.nvidia.com/cuda-downloads).
+- Install the [CUDA 11.0 toolkit and above](https://developer.nvidia.com/cuda-downloads).
 

nvJPEG/nvJPEG-Decoder-Backend-ROI/CMakeLists.txt

@@ -0,0 +1,46 @@
+# 
+# Copyright (c) 2021, NVIDIA CORPORATION.  All rights reserved.
+# 
+# NVIDIA CORPORATION and its licensors retain all intellectual property
+# and proprietary rights in and to this software, related documentation
+# and any modifications thereto. Any use, reproduction, disclosure or
+# distribution of this software and related documentation without an express
+# license agreement from NVIDIA CORPORATION is strictly prohibited.
+# 
+
+cmake_minimum_required(VERSION 3.13 FATAL_ERROR)
+
+project(nvJPEGROIDecode LANGUAGES CXX CUDA)
+
+# ---[ Project specIFication.
+SET(PROJECT_NAME nvJPEGROIDecode)
+PROJECT(${PROJECT_NAME} LANGUAGES CUDA CXX)
+
+if(NOT DEFINED CMAKE_CUDA_STANDARD)
+    set(CMAKE_CUDA_STANDARD 11)
+    set(CMAKE_CUDA_STANDARD_REQUIRED ON)    
+endif()
+
+set(CMAKE_CXX_STANDARD 11)
+set(CMAKE_CXX_STANDARD_REQUIRED ON)
+set(CMAKE_CXX_EXTENSIONS OFF)
+
+
+include_directories(
+  SYSTEM ${CMAKE_CUDA_TOOLKIT_INCLUDE_DIRECTORIES}
+)
+
+
+SET(EXAMPLES_DESCRIPTOR_SOURCES "nvJPEGROIDecode.cpp")
+
+add_executable(nvJPEGROIDecode ${EXAMPLES_DESCRIPTOR_SOURCES})
+
+find_library(NVJPEG_LIB
+    NAMES nvjpeg
+    PATHS ${CMAKE_CUDA_IMPLICIT_LINK_DIRECTORIES})
+
+find_library(CUDART_LIB 
+    NAMES cudart
+    PATHS ${CMAKE_CUDA_IMPLICIT_LINK_DIRECTORIES})
+
+target_link_libraries(nvJPEGROIDecode PUBLIC ${NVJPEG_LIB} ${CUDART_LIB} pthread)

nvJPEG/nvJPEG-Decoder-Backend-ROI/README.md

@@ -0,0 +1,106 @@
+# nvJPEG-Decoder-DifferentBackEnd
+
+## Description
+The sample codes shows the usage of decouple nvJPEG API on decoding JPEG images. Extra functionalities including different backends and ROI are supported.
+
+## Key Concepts
+Image Decoding, different backends from NVJPEG Library
+
+## Supported SM Architecture 
+SM 3.0 SM 3.5 SM 3.7 SM 5.0 SM 5.2 SM 6.0 SM 6.1 SM 7.0 SM 7.2 SM 7.5 8.0
+
+## Supported OS
+Linux, Windows
+
+## Supported CPU Architecture
+x86_64
+
+## CUDA APIs involved
+NVJPEG
+
+# Building
+Build command
+```
+$ mkdir build
+$ cd build
+$ cmake ..
+$ make
+```
+
+# Usage
+```
+Usage: ./nvJPEGROIDecode -i images_dir [-roi roi_regions] [-backend backend_enum] [-b batch_size] [-t total_images] [-w warmup_iterations] [-o output_dir] [-pipelined] [-batched] [-fmt output_format]
+Parameters: 
+        images_dir      :       Path to single image or directory of images
+        roi_regions     :       Specify the ROI in the following format [x_offset, y_offset, roi_width, roi_height]
+        backend_eum     :       Type of backend for the nvJPEG (0 - NVJPEG_BACKEND_DEFAULT, 1 - NVJPEG_BACKEND_HYBRID,
+                                2 - NVJPEG_BACKEND_GPU_HYBRID)
+        batch_size      :       Decode images from input by batches of specified size
+        total_images    :       Decode this much images, if there are less images 
+                                        in the input than total images, decoder will loop over the input
+        warmup_iterations       :       Run this amount of batches first without measuring performance
+        output_dir      :       Write decoded images as BMPs to this directory
+        pipelined       :       Use decoding in phases
+        batched         :       Use batched interface
+        output_format   :       nvJPEG output format for decoding. One of [rgb, rgbi, bgr, bgri, yuv, y, unchanged]
+```
+
+# Example 1 - Choosing different backend
+**Default Backend**
+<br />
+Command: ./nvJPEGROIDecode -i ../input_images/ -backend 0
+```
+...
+...
+Total decoding time: 17.238
+Avg decoding time per image: 1.4365
+Avg images per sec: 0.696136
+Avg decoding time per batch: 1.4365
+```
+
+**Hybrid BackEnd**
+<br />
+Command: ./nvJPEGROIDecode -i ../input_images/ -backend 1
+```
+...
+...
+Total decoding time: 13.9479
+Avg decoding time per image: 1.16233
+Avg images per sec: 0.860344
+Avg decoding time per batch: 1.16233
+```
+
+**GPU Backend**
+<br />
+Command: ./nvJPEGROIDecode -i ../input_images/ -backend 2
+```
+...
+...
+Total decoding time: 20.5875
+Avg decoding time per image: 1.71563
+Avg images per sec: 0.582877
+Avg decoding time per batch: 1.71563
+```
+
+# Example 2 - Decode with ROI
+Command: ./nvJPEGROIDecode -i ../input_images/ -roi 0,0,64,64
+
+Original image: 
+<br />
+![Original Image](./input_images/img9.jpg)
+<br />
+<br />
+Result: 
+<br />
+![Result](img9_roi.png)
+
+
+
+
+
+
+
+
+
+
+