Sprint Goal Setting

Sprint goals 15/10/2022

• Look at the format of the paper
• Look at the BOP dataset challenges
• Write a mid-term report

Sprint goals 15/09/2022

• Start with one research question and experiment on the same.
• Write a report of the first RQ1

Sprint goals 15/08/2022

• Evaluation with Gaussian process, result, and summary of the results
• Datalaoder for three datasets, test case for loading and visualizing
• Train on all the data, results
• COnfusion matrix
• How to compare single-frame segmentation (Vanilla) vs. multi-frame segmentation (Latent frame fusion, GP)
• Visualize the output
• Extended goals: Experiment with two dataset
• Improve the code structure and convert to .py and import it.

Sprint goals 15/07/2022

• Find one dataset
• Focus on the first research question
• Find a list of datasets for semantic segmentation with the camera dataset (Look for Synthetic dataset)
• Evaluation of the model with and without the Gaussian process
• Create a table of results

Sprint goals

MVS2 - Deadline - 15/06/2022

• ⏳ Understand the importance of the Gaussian process on the depth estimation
• ⏳ Literature review on the multi-view stereo
• • Reproduce the result from the paper
• ⏳ Load the dataset and make it run
• ☑️ Download the dataset
• • Understand the differences between MVS approach and MVDepthnet
• • Different methods to construct the cost volume
• • Create a baseline

MVS1 - Deadline - 15/05/2022

• ☑️ Deploy the pretrained model on the android phone
• ⏳ Understand the importance of the Gaussian process on the depth estimation
• ⏳ Literature review on the multi-view stereo
• ☑️ Create a markdown file explaining the deployment of the PyTorch model on the android using Python and Kotlin code
• ☑️ Deploying python code on android
• • Reproduce the result from the paper
• ⏳ Load the dataset and make it run

Points

• Able to deploy on the android device
• Proper setup of system for working
• More focus on the literature work
• Unable to download all the data
• Creating a baseline
• Setting up the research questions
• Time for preparing the slides for presentation with the professor