안녕하세요😄

2024 KHUGGLE 대회에 참가해주신 여러분을 환영합니다. 

다음과 같이 대회에 대해 안내 드립니다.

※ 대회 목표
1.  주어진 데이터 셋을 가장 범용적으로 잘 학습하는 INR(Implicit Neural Representation)  구조를 설계
2. INR 구조를 빠르면서 좋은 성능을 보이도록 하는 학습 방법을 개발
3. Metric과 Competition Rule에 알맞는 범위 안에서 높은 성능 달성 

=================================================================
※ 대회 일정
시작 : 2024.12.06 13:00 
종료 : 2024.12.17 23:59
검토 : 2024.12.18 ~ 2024.12.20(Private score +  Reproduce)
수상 : 2024.12.23 ~ 27 예정

=================================================================
※ 수상 정보
![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F9683477%2F1ceb001ef85d11c9879977832411bfb6%2F.png?generation=1733380370073074&alt=media)

=================================================================
※ 대회 규칙
1. INR 이미지당 Test Step 100회 제한(Train은 상관없음)
2. INR 모델은 각 이미지에 동일한 아키텍쳐를 가져야함(이미지별로 다른 모델 사용 불가)
3. Google Colab에서 실행 가능해야함. (종료 후 Ipynb로 제출 및 재현)
4. 대회 종료 후 Reproduce 불가 및 성능 재현 안될 시 등수 무효
5. 하루 15회 제출 제한


=================================================================
Hello! 😄

Welcome to the 2024 KHUGGLE Competition!

Here is some information about the competition:

※ Competition Objectives
1. Design an INR (Implicit Neural Representation) structure that can generalize and learn the given dataset effectively.
2. Develop training methods that enable the INR structure to achieve fast and high performance.
3. Achieve high performance within the scope of the provided metrics and competition rules.

=================================================================

※  Competition Schedule
Start: December 6, 2024, 13:00
End: December 17, 2024, 23:59
Review Period: December 18, 2024 - December 20, 2024 (Private score +  Reproduce)
Awards Announcement: Scheduled for December 23 - December 27, 2024

=================================================================
※ Competition Rules
1. The INR model is limited to 100 test steps per image (no restrictions on training steps).
2. The INR model must use the same architecture for all images (different models for individual images are not allowed)
3. The solution must be executable on Google Colab (final submission should include a reproducible .ipynb file).
4. Reproducibility: Rankings will be invalidated if the results cannot be reproduced or the performance cannot be replicated after the competition ends.
5. You restricted to submit only 15 per day.

※ Contact Information
Mobile : 010-9383-2203
E-mail : jrespect.im@khu.ac.kr
Office : Electronics Building 350

Thank you for your participation, and good luck to all contestants! 🎉