# Bring Your Own Project for 2y (2024-25)

Hello 2Y! The much anticipated BYOP is here a little earlier than usual. For those unaware, this is DSG's annual event to recruit enthusiastic and talented sophomores. So, let's dive right in!

## Timeline

- 14/10/2024: **Proposal submission begins.**
- 20/10/2024: **Proposal submission ends.**
- 24/10/2024 **[UPDATED]**: **Announcement of proposal selection results.**
- 25/10/2024 **[UPDATED]**: **Phase 1 starts**
- 07/11/2024 **[UPDATED]**: **MidEval, i.e. Phase 1 ends**
- 08/11/2024 to 15/11/2024: **ETE Prep time**
- 16/11/2024 to 27/11/2024: **ETE**
- 28/11/2024: **Phase 2 starts**
- 13/12/2024: **Final Submission**

## Tracks

Following a successful recruitment season last year, we continue with the **Open Track** and **Reproducibility Track**.

### Open Track:

You know the drill if you participated in BYOP during your first year! You have to come up with any ML related project and submit a detailed proposal.

[Sample Proposal](https://docs.google.com/document/d/1_QomDggpHixLHNkVFW42xJL3VKFzorhsSSOy2-fUkgY/edit?usp=sharing)

**Note:** 2Y proposals must be more rigorous and detailed than you did in your 1st years. It should provide a clear roadmap and attach all relevant papers and resources you referred to or plan to use. Similarly, projects need to be challenging enough for a sophomore. We encourage you to contact us with your proposal before submission to ensure your projects are good enough.

### Reproducibility Track:
Inspired by the Machine Learning Reproducibility Challenge, conducted annually, we continue with the reproducibility track. For more details about the challenge, go through [this blog post.](https://reproml.org/blog/announcing_mlrc2023/)

**Please go through the blog post above.**
**tl;dr:**
- Choose a paper from top conferences (CVPR, ECCV, ICCV, NeurIPS, ICML, ICLR, etc). It is recommended that the paper be recent i.e. published in 2024.
- Reproduce its results and possibly conduct ablation studies or extra experiments to verify its claims and come up with novel insights.
- Choice of paper should be based on the availability of code, data and scope of extra experiments.
- Make a proposal following the sample proposal below and submit it.

[Sample Proposal](https://docs.google.com/document/d/1AndDjbo2YuRgYZAe-1UCDLggW0QMKSftoTvsiXZrVH8/edit?tab=t.0)

### Guidelines:
* The end deliverables of this project will be a GitHub repo with a description containing the code base (preferably in the readme files).
* Your proposal should contain clarity about your working goals until mid and end evaluation.
* Answer every section in the sample proposal in detail and strictly follow the format.
* Brownie points if a demo of the result is deployed.

**Selected Proposals from last year for reference:**
- [Open Track](https://docs.google.com/document/d/1x67loRWbDHxK89dHFtXEBHP1qNFBc6DSBM27DU_kiMs/edit?tab=t.0)
- [Reproducibility Track](https://drive.google.com/file/d/1oetz4doh2jMdGYxZDUCHP6Al2fNJXoev/view)

## Submission

[Submission Form](https://docs.google.com/forms/d/e/1FAIpQLSd4DcyPDhoR9DGp3cK7AXWgr3a-fjSMN-r_aA_YSMUcwfi0Tw/viewform?usp=sf_link)