Project Deliverables for Denoising and Training

1. Visualization of the Noising Process with Steps

This section demonstrates the gradual addition of noise to an image at different time steps, visualizing the noising process.

2. Training Loss Curve Recorded Periodically

A plot of the training loss recorded every few iterations during the whole training process.

3. Results from Test Set After First Epoch

Sample results from the test set after training the model for one epoch, showcasing early performance.

Sample results from the test set after training the model for five epochs.

4. Results for Out-of-Distribution Noise Levels

Sample results with varying noise levels outside the range seen during training. Same image used for consistency.

5. Full Training Loss Curve for Time-Conditioned UNet

A plot showing the loss over the entire training process for the time-conditioned UNet model.

6. Sample Outputs for Time-Conditioned UNet

Generated samples after training the time-conditioned UNet for 5 and 20 epochs, highlighting improvements over time.

7. Full Training Loss Curve for Class-Conditioned UNet

A plot showing the loss over the entire training process for the class-conditioned UNet model.

8. Outputs from Class-Conditioned UNet for Multiple Digits

Generated results for class-conditioned UNet after 5 and 20 epochs, including four variations for each digit.