AlexNet-MNIST-Inference

Short description

A lightweight C++ library for performing high-performance inference on MNIST handwritten digits using a modified AlexNet architecture. Designed for efficiency and educational purposes, this project demonstrates how classic CNNs can be optimized for small-scale tasks in native environments.

Key Features:

• C++17 implementation for bare-metal performance

• Simplified AlexNet for 28×28 grayscale images

• Parallel computing via Intel OneTBB (Threading Building Blocks)

• Pre-trained model: AlexNet-model.h5 included

Some files used to create the library

Neural network models

You need to download Alexnet-model.h5 to the folder docs

How do I launch the inference?

• Make sure you install the project dependencies by running: pip install -r requirements.txt
• You need to run the script parser.py that is located in app/AlexNet to read weights from a model Alexnet-model.h5 and the json file with the weights will be stored in the docs folder.
• Then put the test images in png format in the folder docs/input
• After building the project, which is described below, run Graph_build in folder build/bin

Building a Project

Windows

To build and run this project locally on Windows, follow these steps:

1. Clone the repository:
   Clone this repository to your local machine using the following command:

   git clone https://github.com/embedded-dev-research/itlab_2023.git

2. Update submodules: Navigate to the project directory and update the submodules:

   git submodule update --init --recursive

3. Configure the project: Create a separate directory for configure the project and compile it:

   mkdir build
   cd build
   cmake .. -DCMAKE_BUILD_TYPE=Release

   If you want to build in a Debug, change the Release to Debug

   Note: Make sure you have CMake installed to build the project.

4. Build the project: Next, to build the project, we will need to enter the command

   cmake --build . --config Release

5. Run the project After building the project, you can find the executable file in the following path from the build folder

   cd build/bin

   and run the file

   Graph_Build.exe

Linux/macOS

To build and run this project locally on Linux or macOS, follow these steps:

1. Clone the repository:
   Clone this repository to your local machine using the following command:

   git clone https://github.com/embedded-dev-research/itlab_2023.git

2. Update submodules: Navigate to the project directory and update the submodules:

   git submodule update --init --recursive

3. Install necessary dependencies:
   1. OpenMP Debian/Ubuntu:

   sudo apt-get install -y libomp-dev

   macOS:

   brew install libomp
   

4. Configure the project: Create a separate directory for configure the project and compile it:

   cmake -S . -B build

   Note: Make sure you have CMake installed to build the project. For macOS need to specify the path to omp.h file:

   cmake -S . -B build -DCMAKE_CXX_FLAGS="-I$(brew --prefix libomp)/include" -DCMAKE_C_FLAGS="-I$(brew --prefix libomp)/include"

5. Build the project: Next, to build the project, we will need to enter the command

   cmake --build build --config Release

   If you want to build in a Debug, change the Release to Debug
6. Run the project After building the project, you can find the executable file in the following path from the build folder

   cd build/bin

   and run the file

   ./Graph_Build

Test Process

This project contains tests to verify functionality. To test the project, the Google Test Framework is used as a submodule of the project.

Google Test Framework

Google Test is a powerful framework for unit testing in C++. In this project, Google Test is a submodule. When building the project, you have already updated it, and it is ready for use.

Running tests

Windows

To start the testing process locally, you need to go to the directory

cd build/bin

and run the following files:

run_test.exe

Linux

To start the testing process locally, you need to go to the directory

cd build/bin

and run the following files:

chmod +x run_test
./run_test

Accuracy validation

To run accuracy validation you need to use the MNIST dataset, which you can download here and put it in a folder docs/mnist/mnist/test Now you can run accuracy check - build\bin\ACC_MNIST.exe

• The accuracy should be 98.02%

Documentation of project

https://github.com/embedded-dev-research/itlab_2023/blob/Semyon1104/Final_documentation/docs/IT_Lab_2023.pdf

Structure of our library