Introduction

1. Gather and Label Training Images

This step is crucial for ensuring the accuracy of the model. The training images should feature random objects alongside the desired ones, and they should encompass a diverse range of backgrounds and lighting conditions. For this project, each label will consist of 50 different images. With four distinct Turkish Sign Language words in mind, I will gather a total of 200 images and label them using LabelImg.

2. Install TensorFlow Object Detection Dependencies

I chose to install the TensorFlow Object Detection API in Google Colab for its convenience. This involves cloning the TensorFlow models repository and executing a couple of installation commands. You can access the code segment in the Google Colab instance where I trained the model.

3. Upload Image Dataset and Prepare Training Data

In this step, we will upload our images, organize them into train, validation, and test folders, and convert them into TFRecords format for training. First, on your local PC, zip all your training images and XML files into a single folder called "images.zip". The files should be directly inside the zip folder, or in a nested folder as shown below:

4. Set Up Training Configuration

In this section, we'll set up the model and training configuration. We'll specifiy which pretrained TensorFlow model we want to use from the TensorFlow 2 Object Detection Model Zoo. For this project I used ssd-mobilenet-v2-fpnlite-320.

5. Train Custom TFLite Detection Model

This is the PLACE where we struggle the most. We're ready to train our object detection model! Model training is performed using the "model_main_tf2.py" script from the TF Object Detection API. Training will take anywhere from 2 to 6 hours, depending on the model, batch size, and number of training steps.

6. Convert Model to TensorFlow Lite

Yuppiee! Our model is all trained up and ready to be used for detecting objects.

7. Test TensorFlow Lite Model and Calculate mAP

We've trained our custom model and converted it to TFLite format. But how well does it actually perform at detecting objects in images? This is where the images we set aside in the test folder come in. The model never saw any test images during training, so its performance on these images should be representative of how it will perform on new images from the field.

8. Deploy TensorFlow Lite Model

Now that your custom model has been trained and converted to TFLite format, it's ready to be downloaded and deployed in an application!

Here are links to the deployment guides for Windows, Linux, and macOS:

• How to Run TensorFlow Lite Models on Windows
• How to Run TensorFlow Lite Models on macOS