Loading [MathJax]/extensions/tex2jax.js
MSE-CNN Implementation 1
Code database with the implementation of MSE-CNN, from the paper 'DeepQTMT: A Deep Learning Approach for Fast QTMT-based CU Partition of Intra-mode VVC'
All Classes Namespaces Files Functions Variables Pages
Example data

The data in this folder is used by the scripts in the src/msecnn_raulkviana/usefull_scripts folder. The data in this folder is unbalanced. To load this data into the custom data class, you can follow this sample of code

import custom_dataset
# Path to already processed data
l_path = "path_to_label"
# Instatiate dataset class to train stage 6 (there is also classes to train other stages)
dataset = custom_dataset.CUDatasetStg6(files_path=l_path_train) # Dataset for training of stage 6
batch_sampler = custom_dataset.SamplerStg6(train_data, batch_size) # Batch Sampler to batch cus of the same size together
dataloader_train = DataLoader(train_data, num_workers=num_workers, batch_sampler=batch_sampler_train) # Data Loader that can be used to train or evaluate the model