- Install tensorflow, tf-slim and other dependencies required by the code (Ideally in a virtual environment).
- Download VGG model. http://download.tensorflow.org/models/vgg_16_2016_08_28.tar.gz
- Untar the model and then update the checkpoints_dir variable as folder containing the extracted vgg model
- Update log folder variable log_folder
- Lines 272 - 279 have been temporarily commented to prevent plots being showed after each iteration. Plots holds up further execution till the plot window is closed.
- The input image size should be 480x352. All images must be resized to this dimension before being fed to the algorithm.
- Replace vaiable named 'processed_probabilities' with 'softmax'(Already done in the code).
senyao-han / crf-image-segmentation Goto Github PK
View Code? Open in Web Editor NEWThis project forked from therealnidhin/crf-image-segmentation