Topic recognition#295

Open

GuanhuaMa wants to merge 71 commits intoshakes76:topic-recognitionfrom

GuanhuaMa:topic-recognition

Commits on Oct 25, 2025

Add 1_UNet_48339261 project folder
GuanhuaMa
committed

Commits on Oct 26, 2025

Add empty Python files for 1_UNet_48339261
GuanhuaMa
committed

Commits on Oct 28, 2025

feat: Initialise HipMRIDataset class structure
GuanhuaMa
committed
feat: Implement file searching logic in __init__
GuanhuaMa
committed
feat: Convert loaded numpy arrays to PyTorch tensors
GuanhuaMa
committed
feat: Add Z-score normalization for image tensors
GuanhuaMa
committed
feat: Implement binary mask for prostate label
GuanhuaMa
committed
fix: Add channel dimension to tensors
GuanhuaMa
committed
fix: Squeeze mask shape to [H, W] for loss function
GuanhuaMa
committed
feat: Add resize for various image sizes
GuanhuaMa
committed
fix: import tourch library
GuanhuaMa
committed
feat: Add initial training function
GuanhuaMa
committed
feat: Add batch loop
GuanhuaMa
committed
feat: Implement forward pass and loss calculation
GuanhuaMa
committed
feat: Add optimizer and backpropagation
GuanhuaMa
committed
feat: Add loss tracking and epoch logging
GuanhuaMa
committed
feat: Add a Visualize predictions after each epoch
GuanhuaMa
committed
fix: change a variable name - prediction_squeezed
GuanhuaMa
committed
feat: Add device to use gpu or cpu
GuanhuaMa
committed
feat: Add dice_score metric for segmentation evaluation
GuanhuaMa
committed
feat: Add plot train loss curve function
GuanhuaMa
committed
feat: Add prediction visualization function
GuanhuaMa
committed
fix: add model.eval() and no_grad()
GuanhuaMa
committed
fix: use calculate_dice_score into prediction plot
GuanhuaMa
committed
feat: using .copy() to prevent the negative stride
GuanhuaMa
committed
feat: import library
GuanhuaMa
committed
fix: using plt.savefig inseat of plt.show
GuanhuaMa
committed
feat: import library and implement if_main function
GuanhuaMa
committed
feat: implement show_prediction function
GuanhuaMa
committed
chore: mark restore of predict.py (no content change)
GuanhuaMa
committed
feat: Model loading and Dataset loading
GuanhuaMa
committed
feat: add visualization
GuanhuaMa
committed
feat: implement if_main with parameters and dataset loading
GuanhuaMa
committed
feat: add training process in if_main
GuanhuaMa
committed
fix: delete a 'pass'
GuanhuaMa
committed
feat: Add SimpleUNet with Conv-ReLU blocks
GuanhuaMa
committed
feat: Implement DiceLoss function
GuanhuaMa
committed
feat: Add BatchNorm2d to conv blocks
GuanhuaMa
committed
feat: Replace ReLU with LeakyReLU
GuanhuaMa
committed
feat: Add Dropout2d for regularization
GuanhuaMa
committed
feat: Move Sigmoid to Model output
GuanhuaMa
committed
fix: use reshape instead of view in DiceLoss
GuanhuaMa
committed
fix: add dropout_p parameter in function definition
GuanhuaMa
committed
fix: add random_split library
GuanhuaMa
committed
Rename project folder to UNet_task3_48339261
GuanhuaMa
committed

Commits on Oct 30, 2025

fix: update data loading to match real dataset folder structure
GuanhuaMa
committed
Add .gitignore to exclude .DS_Store
GuanhuaMa
committed
fix: delete repetitive code
GuanhuaMa
committed
fix: update the dataset path and add a checking for file exist
GuanhuaMa
committed
fix: delete the random split and use the new split method
GuanhuaMa
committed
fix: change the Inappropriate word
GuanhuaMa
committed
change the partition zone
GuanhuaMa
committed
Rename run_task3.slurm to job.slurm
GuanhuaMa
committed
fix: change the data path for process in Colab
GuanhuaMa
committed
fix: change the data path for process in Colab
GuanhuaMa
committed
write report in it
GuanhuaMa
committed
write report in it
GuanhuaMa
committed
docs: Remove Rangpur .slurm file, useless file
GuanhuaMa
committed
docs: Add Run_on_Colab.ipynb for reproducibility
GuanhuaMa
committed
fix: specify the process for reproducibility on Colab
GuanhuaMa
committed
cite: the code SimpleUNet in UNet_segmentation_code_demo.ipynb
GuanhuaMa
committed
cite: Ai assistant for stride error and mask resize problem
GuanhuaMa
committed
cite: Ai assistant
GuanhuaMa
committed
fix: specify the process for reproducibility on Colab
GuanhuaMa
committed
Add Figure 1 & 2 to show in README.md
GuanhuaMa
authored
Fix image links in README.md
GuanhuaMa
authored
Rename final_predictions (1).png to final_predictions_(1).png
GuanhuaMa
authored
Rename training_loss_curve (1).png to training_loss_curve.png
GuanhuaMa
authored
Rename final_predictions_(1).png to final_predictions.png
GuanhuaMa
authored
Fix image links in README.md
GuanhuaMa
authored
Fix image markdown for training loss curve
GuanhuaMa
authored