TensorFlow Object Detection API tutorial

Important

This tutorial is intended for TensorFlow 1.14, which (at the time of writing this tutorial) is the latest stable version before TensorFlow 2.x.

Tensorflow 1.15 has also been released, but seems to be exhibiting instability issues.

A version for Tensorflow 1.9 can be found here.

A version for Tensorflow 2.x is in the making and a link will be added here when ready.

This is a step-by-step tutorial/guide to setting up and using TensorFlow’s Object Detection API to perform, namely, object detection in images/video.

The software tools which we shall use throughout this tutorial are listed in the table below:

Target Software versions
OS	Windows, Linux
Python	3.7
TensorFlow	1.14
CUDA Toolkit	10.0
CuDNN	7.6.5
Anaconda	Python 3.7 (Optional)

Contents:

• Installation
  • General Remarks
  • Install Anaconda Python 3.7 (Optional)
  • TensorFlow Installation
    • TensorFlow CPU
      • Create a new Conda virtual environment (Optional)
      • Install TensorFlow CPU for Python
      • Test your Installation
    • TensorFlow GPU
      • Install CUDA Toolkit
      • Install CUDNN
      • Environment Setup
      • Update your GPU drivers (Optional)
      • Create a new Conda virtual environment
      • Install TensorFlow GPU for Python
      • Test your Installation
  • TensorFlow Models Installation
    • Install Prerequisites
    • Downloading the TensorFlow Models
    • Protobuf Installation/Compilation
    • Adding necessary Environment Variables
    • COCO API installation (Optional)
    • Test your Installation
  • LabelImg Installation
    • Get from PyPI (Recommended)
    • Use precompiled binaries (Easy)
    • Build from source (Hard)
• Detect Objects Using Your Webcam
• Training Custom Object Detector
  • Preparing workspace
  • Annotating images
  • Partitioning the images
  • Creating Label Map
  • Creating TensorFlow Records
    • Converting *.xml to *.csv
    • Converting from *.csv to *.record
  • Configuring a Training Pipeline
  • Training the Model
  • Evaluating the Model (Optional)
  • Monitor Training Job Progress using TensorBoard
  • Exporting a Trained Inference Graph
• Common issues
  • Python crashes - TensorFlow GPU
  • Cleaning up Nvidia containers (TensorFlow GPU)
  • labelImg saves annotation files with .xml.xml extension
  • “WARNING:tensorflow:Entity <bound method X of <Y>> could not be transformed …”

Indices and tables

• Index

• Module Index

• Search Page