Semantic segmentation using CNNs, performed on public datasets like Oxford/IIIT Pets and PASCAL VOC
Read about Image Segmentation here (where the modded_unet model was taken from): https://www.tensorflow.org/tutorials/images/segmentation
Tensorflow v2.3.1
Tensorboard v2.3
Numpy v1.18.5
Matplotlib v3.3.3
Tensorflow datasets (for Oxford/IIIT Pets dataset) (pip install tfds-nightly
) v4.1.0
-
Use the shell script at https://github.com/tensorflow/models/blob/master/research/deeplab/datasets/download_and_convert_voc2012.sh or download the datasets directly from http://host.robots.ox.ac.uk/pascal/VOC/voc2012/ and build the TFRecords with https://github.com/tensorflow/models/blob/master/research/deeplab/datasets/build_voc2012_data.py
-
Save the TFRecord file at ./datasets/ (Unable to upload the original files due to Github's 25MB file upload limit