imageNet.h

Go to the documentation of this file.

/*
 * Copyright (c) 2017, NVIDIA CORPORATION. All rights reserved.
 *
 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the "Software"),
 * to deal in the Software without restriction, including without limitation
 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
 * and/or sell copies of the Software, and to permit persons to whom the
 * Software is furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in
 * all copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
 * DEALINGS IN THE SOFTWARE.
 */
 
#ifndef __IMAGE_NET_H__
#define __IMAGE_NET_H__
 
 
#include "tensorNet.h"
 
 
#define IMAGENET_DEFAULT_INPUT   "data"
 
#define IMAGENET_DEFAULT_OUTPUT  "prob"
 
#define IMAGENET_DEFAULT_THRESHOLD 0.01f
 
#define IMAGENET_MODEL_TYPE "classification"
 
#define IMAGENET_USAGE_STRING  "imageNet arguments: \n"                                                         \
                  "  --network=NETWORK    pre-trained model to load, one of the following:\n"   \
                  "                           * alexnet\n"                                                              \
                  "                           * googlenet (default)\n"                                  \
                  "                           * googlenet-12\n"                                                         \
                  "                           * resnet-18\n"                                                    \
                  "                           * resnet-50\n"                                                    \
                  "                           * resnet-101\n"                                                   \
                  "                           * resnet-152\n"                                                   \
                  "                           * vgg-16\n"                                                               \
                  "                           * vgg-19\n"                                                               \
                  "                           * inception-v4\n"                                                         \
                  "  --model=MODEL        path to custom model to load (caffemodel, uff, or onnx)\n"                    \
                  "  --prototxt=PROTOTXT  path to custom prototxt to load (for .caffemodel only)\n"                             \
                  "  --labels=LABELS      path to text file containing the labels for each class\n"                             \
                  "  --input-blob=INPUT   name of the input layer (default is '" IMAGENET_DEFAULT_INPUT "')\n"  \
                  "  --output-blob=OUTPUT name of the output layer (default is '" IMAGENET_DEFAULT_OUTPUT "')\n"        \
                  "  --threshold=CONF     minimum confidence threshold for classification (default is 0.01)\n"  \
                  "  --smoothing=WEIGHT   weight between [0,1] or number of frames (disabled by default)\n"             \
                  "  --profile            enable layer profiling in TensorRT\n\n"
 
 
class imageNet : public tensorNet
{
public:
        typedef std::vector<std::pair<uint32_t, float>> Classifications;
        
        static imageNet* Create( const char* network="googlenet", 
                                                uint32_t maxBatchSize=DEFAULT_MAX_BATCH_SIZE, 
                                                precisionType precision=TYPE_FASTEST,
                                                deviceType device=DEVICE_GPU, bool allowGPUFallback=true );
                                                
        static imageNet* Create( const char* prototxt_path, const char* model_path, 
                                                const char* mean_binary, const char* class_labels, 
                                                const char* input=IMAGENET_DEFAULT_INPUT, 
                                                const char* output=IMAGENET_DEFAULT_OUTPUT, 
                                                uint32_t maxBatchSize=DEFAULT_MAX_BATCH_SIZE, 
                                                precisionType precision=TYPE_FASTEST,
                                                deviceType device=DEVICE_GPU, bool allowGPUFallback=true );
        
        static imageNet* Create( int argc, char** argv );
 
        static imageNet* Create( const commandLine& cmdLine );
 
        static inline const char* Usage()               { return IMAGENET_USAGE_STRING; }
 
        virtual ~imageNet();
                        
        template<typename T> int Classify( T* image, uint32_t width, uint32_t height, float* confidence=NULL )          { return Classify((void*)image, width, height, imageFormatFromType<T>(), confidence); }
        
        int Classify( void* image, uint32_t width, uint32_t height, imageFormat format, float* confidence=NULL );
 
        int Classify( float* rgba, uint32_t width, uint32_t height, float* confidence=NULL, imageFormat format=IMAGE_RGBA32F );
 
        template<typename T> int Classify( T* image, uint32_t width, uint32_t height, Classifications& classifications, int topK=1 )            { return Classify((void*)image, width, height, imageFormatFromType<T>(), classifications, topK); }
 
        int Classify( void* image, uint32_t width, uint32_t height, imageFormat format, Classifications& classifications, int topK=1 );
 
        inline uint32_t GetNumClasses() const                                           { return mNumClasses; }
        
        inline const char* GetClassLabel( int index ) const                     { return GetClassDesc(index); }
        
        inline const char* GetClassDesc( int index )    const                   { return index >= 0 ? mClassDesc[index].c_str() : "none"; }
        
        inline const char* GetClassSynset( int index ) const                    { return index >= 0 ? mClassSynset[index].c_str() : "none"; }
        
        inline const char* GetClassPath() const                                         { return mClassPath.c_str(); }
        
        inline float GetThreshold() const                                                       { return mThreshold; }
        
        inline void SetThreshold( float threshold )                                     { mThreshold = threshold; }
        
        inline float GetSmoothing() const                                                       { return mSmoothingFactor; }
        
        inline void SetSmoothing( float factor )                                        { mSmoothingFactor = factor; }
 
protected:
        imageNet();
        
        //bool init( NetworkType networkType, uint32_t maxBatchSize, precisionType precision, deviceType device, bool allowGPUFallback );
        bool init(const char* prototxt_path, const char* model_path, const char* mean_binary, const char* class_path, const char* input, const char* output, uint32_t maxBatchSize, precisionType precision, deviceType device, bool allowGPUFallback );
        bool loadClassInfo( const char* filename, int expectedClasses=-1 );
        
        bool preProcess( void* image, uint32_t width, uint32_t height, imageFormat format );
        
        float* applySmoothing();
        
        uint32_t mNumClasses;
        
        std::vector<std::string> mClassSynset;  // 1000 class ID's (ie n01580077, n04325704)
        std::vector<std::string> mClassDesc;
 
        std::string mClassPath;
        //NetworkType mNetworkType;
        
        float* mSmoothingBuffer;
        float  mSmoothingFactor;
        
        float mThreshold;
};
 
 
#endif