Some code for the Semantic Segmentation in TensorFlow/Keras Data Hub TechTalk, March 7 2023

1. clone this repo
2. download the data (see below) and structure like this:

data
├── class_dict_seg.csv
├── RGB_color_image_masks
├── test_data
├── test_labels
├── train_data
├── train_labels
├── val_data
└── val_labels

The program assumes all images are *.png files that have been resized and cropped to 512x512

3. Create virtual environment (python3 -m venv venv)
4. source ./venv/bin/activate
5. pip install -U pip
6. pip install tensorflow opencv-python scikit-image
7. mkdir raw_output final_output

1. source ./venv/bin/activate
2. python unet.py

1. source ./venv/bin/activate
2. python ./unet_inference.py (inferred output will be in "raw_output")

1. source ./venv/bin/activate
2. python ./convert.py (files will be in "final_output")

https://www.kaggle.com/datasets/bulentsiyah/semantic-drone-dataset

More information on the dataset can be found here: http://dronedataset.icg.tugraz.at/