TI Deep Learning API User’s Guide

• Introduction
  • Key Features
  • Development Flow
• Overview
  • Software Architecture
  • Terminology
• Using the API
  • Use Cases
    • Each EO processes a single frame
    • Frame split across EOs
    • Using EOPs for double buffering
  • Sizing device side heaps
  • Accessing outputs of network layers
• Examples
  • Imagenet
  • Segmentation
  • SSD
  • MNIST
  • Running Examples
    • Image input
    • Camera (live video) input
    • Pre-recorded video (mp4/mov/avi) input
    • Displaying video output
• API Reference
  • Configuration
    • Configuration file
    • Overriding layer group assignment
  • Executor
  • ExecutionObject
  • ExecutionObjectPipeline
• Viewers
  • Network graph viewer
  • Execution Graph viewer
• Frequently Asked Questions
  • Why do I get an assertion failure from malloc_ddr?
    • Allocations not freed
    • Insufficient OpenCL global memory
  • How do I build the TIDL API from source?
• Changelog
  • 1.4.0 [Processor Linux SDK 6.2]
  • 1.3.3 [Processor Linux SDK 6.1]
  • 1.3.0 [Processor Linux SDK 5.3]
  • 1.2.2 [Processor Linux SDK 5.2]
  • 1.1.0 [Processor Linux SDK 5.1]
  • 1.0.0 [Processor Linux SDK 5.0]
• Important Notice
• Disclaimer