Home

Welcome to the Object Detection TensorFlow wiki!

TensorFlow is an Open-source software libraries developed by google to do varies of applications using numerical inputs. It is most used in Machine learning.

This tutorial will work with an existing model MobileNet which contains hundreds of defined objects.

It will be assumed that Python3 is already installed.

To start training your model with tensorflow, some configurations must be configured.

First, Downloads:

1. Anaconda
2. TensorFlow models
3. Repository files

Then, from the search bar type “Anaconda”, the Anaconda Prompt will pop -open it.

Now in Anaconda3 window start dependencies installations:

pip install tensorflow

pip install opencv-python

pip install Cython

pip install contextlib2

pip install pillow

pip install lxml

pip install jupyter

---

Second, move below files from the rasperotiy to models-master/research

• 1.0 Object Detection Tutorial.ipynb
• 1.1 Customized Object Detection - Images.ipynb
• 1.3 Customized Object Detection - Video.ipynb
• protoc.exe

---

Third, now start implementing all together.

In the search bar type “jupyter notebook” open the application -> it will take you to jupyter website window.

On the top bar there is “Upload” button, click it and upload the three files type ipynb from the models/research path.

Then, open the 1.0 Object file. For first time opening, the models will be downloaded automatically from the internet from the Load part in the code.

Go through each part of the code and run it till the end without any error. You can change the photos tested in the code from the test_images folder in models-master/research/object_detection/test_images/ path.

---

You may face an error in importing tensorflow as tf which will be solved if you create an environment for installing the tensorflow, by the commands in the Anaconda3 prompt:

conda create —-name newenv

conda activate newenv

conda install tensorflow