Blueoil – An efficient neural network generator

• 1. Installation
  • 1.1. Prerequisites
  • 1.2. Install / Uninstall Blueoil
  • 1.3. Setup an FPGA Board
  • 1.4. [OPTIONAL] Create your own Linux image on microSD card
  • 1.5. Install required packages
  • 1.6. Other Information
• 2. Getting started tutorials
  • 2.1. Training for Basic Classification
  • 2.2. Training for Object Detection
  • 2.3. Training for Semantic Segmentation
  • 2.4. Training for Keypoint Detection
  • 2.5. Run a trained neural network on the FPGA board
• 3. Usage Guide
  • 3.1. Prepare training dataset
  • 3.2. Generate a configuration file
  • 3.3. Train a neural network
  • 3.4. Convert your training result to FPGA ready format
  • 3.5. Time measurement
• 4. FAQ
  • 4.1. Connection
  • 4.2. OS
  • 4.3. Errors
• 5. Converter Documentation
  • 5.1. Overview
  • 5.2. Current acceptable graph representation
  • 5.3. Installation
  • 5.4. Getting Started
  • 5.5. Custom project generation & Optimizations
  • 5.6. Prepare input directory and data
  • 5.7. Prepare output directory
  • 5.8. Get results
  • 5.9. Shared libraries
  • 5.10. Archive file
  • 5.11. Supported Ops
• 6. Quantize Documentation
  • 6.1. Weight quantization
  • 6.2. Activation quantization
  • 6.3. Simple Quantized network
• 7. Specification
  • 7.1. Network model’s name, version and structure
  • 7.2. Network specification
  • 7.3. Output Data Specification
  • 7.4. Format Specification
• 8. API Reference
  • 8.1. blueoil