OBSCURA | OFFICIAL PARTICIPATION AND SUBMISSION RULES

Important: These rules apply together with Devpost's Terms of Service, Kaggle's Foundational Competition Rules, and the official OBSCURA competition-specific rules published on Kaggle.

By participating in OBSCURA, participants agree to follow the applicable eligibility, submission, data-use, conduct, and platform requirements. If any requirements conflict, the applicable platform's binding rules and competition-specific terms must be respected.

01 | STUDENT ELIGIBILITY

• OBSCURA is a student-led research competition open exclusively to currently enrolled students at recognized colleges and universities.
• Eligible students from different institutions and countries may participate, subject to applicable age, geographic, residency, sanctions, and platform restrictions.
• Participants must meet Kaggle's eligibility requirements, including its minimum age restrictions.
• Organizers may request proof of student enrollment, identity, or other relevant eligibility information before awarding prizes or certificates.
• Organizers and individuals who developed the competition dataset, prepared the hidden targets, or had access to private evaluation answers are not eligible to compete for prizes.

02 | INDIVIDUALS AND TEAMS

• Participants may compete individually or in teams of up to 20 eligible students.
• Students from different recognized colleges or universities may collaborate within the same team.
• Each participant may use only one Kaggle account and belong to only one competition team.
• Every team member must satisfy the student eligibility requirements.
• The Devpost project must identify the same participants as the corresponding Kaggle entry.
• Team formation, mergers, membership changes, and final submission selection must follow Kaggle's published competition rules and deadlines.
• Any team exceeding the maximum permitted size will not be eligible for awards.

03 | REQUIRED SUBMISSIONS ON BOTH PLATFORMS

OBSCURA uses Kaggle for technical model evaluation and Devpost for research documentation and project submissions.

Part A: Kaggle Submission

1. Register for the official OBSCURA competition on Kaggle and accept its rules.
2. Develop a machine learning solution using the provided dataset and permitted resources.
3. Submit at least one valid prediction file following Kaggle's required CSV format.
4. Follow Kaggle's daily submission limits, team requirements, and final submission deadlines.
5. Select final Kaggle submissions according to the official competition rules.

Part B: Devpost Submission

1. Register for the OBSCURA hackathon on Devpost.
2. Create one Devpost project corresponding to your individual Kaggle entry or Kaggle team.
3. Provide a clear project description explaining your research approach and findings.
4. Upload a research write-up in PDF format describing your methodology, experiments, validation, results, and conclusions.
5. Include your Kaggle profile or team information and relevant permitted code or notebook references.
6. Ensure that all required information is submitted before the Devpost deadline.

A participant video presentation is NOT required.

The research write-up PDF is mandatory. Participants may follow the example report provided by the organizers or use their own research report structure, provided all required information is included.

Important: A valid Kaggle submission and a complete Devpost research submission are both required for prize and participation certificate eligibility.

04 | DATA INTEGRITY AND FAIR PLAY

• Do not identify or attempt to reconstruct the real industrial facilities, geographic locations, acquisition dates, or hidden target measurements represented by the competition data.
• Do not use outside satellite, weather, air-quality, methane, fire, elevation, or other prohibited data to identify competition samples or recover their answers.
• Do not manually label test rows or attempt to infer hidden answers through leaderboard probing.
• Do not exploit file metadata, identifiers, row ordering, or other unintended information to gain an unfair advantage.
• Do not create multiple accounts or attempt to bypass Kaggle's submission limits.
• Permitted pretrained models must satisfy the official Kaggle rules and must be disclosed in the research write-up.
• Competition data may only be used for permitted non-commercial, research, educational, and competition-related purposes, subject to the applicable data license and rules.

05 | CODE SHARING, REPRODUCIBILITY, AND ORIGINALITY

• Do not redistribute, mirror, resell, or publicly upload restricted competition datasets, image chips, tables, target values, or identifying sample information outside authorized Kaggle channels.
• Private sharing of competition code, models, predictions, or restricted data between separate teams is prohibited unless explicitly permitted by the official rules.
• Public code sharing must comply with Kaggle's competition rules and must not expose restricted competition data.
• Participants may reference permitted Kaggle notebooks or GitHub repositories in their research PDF.
• All research submissions must truthfully describe the participant's actual experiments, models, methods, and findings.
• Participants must properly acknowledge relevant pretrained models, software libraries, and permitted external resources.
• Potential winners must provide reproducible code and supporting documentation when requested by the organizing team.
• Winning solutions must satisfy the applicable Kaggle code delivery, licensing, originality, and reproducibility requirements before prizes are awarded.

06 | TECHNICAL RANKING, PRIZES, AND CERTIFICATES

Technical Evaluation

Kaggle evaluates prediction submissions using Masked Mean Column-wise Root Mean Squared Error (MCRMSE). Lower scores indicate better performance.

The Kaggle private leaderboard determines final technical ranking. The public leaderboard provides interim feedback and does not determine final awards.

Initial Cash Prize Pool: $50 USD

Cash prizes are awarded per winning entry. An individual winner receives the full prize corresponding to their placement. A winning team receives one prize for the entire team, not a separate cash prize for each member. Distribution among team members is subject to applicable competition rules.

The prize pool may increase if additional sponsors or supporters join. Any confirmed changes will be communicated through the official competition channels.

Prize Eligibility and Verification

• Participants must complete the required Kaggle and Devpost submissions.
• Devpost and Kaggle team membership must match.
• Participants must satisfy the student eligibility requirements.
• Submissions must comply with the competition's technical, conduct, originality, and data-use rules.
• Potential winners must provide reproducible code and any additional documentation required for verification.
• Leaderboard placement alone does not guarantee a cash prize.
• If an entry is disqualified or fails verification, the organizers may consider the next eligible entry in accordance with the official competition rules.

Digital Certificates of Participation

Every eligible student who successfully completes the required submissions on both Kaggle and Devpost will receive an individual digital Certificate of Participation after verification.

Participants do not need to win a cash prize or achieve a particular leaderboard position to receive a participation certificate.

Simply registering for the competition without completing the required submissions does not qualify a participant for a certificate.

07 | COMPETITION TIMELINE

All times are in Pakistan Standard Time (PKT, GMT+5).

Any official schedule changes will be announced through the competition platforms. Participants are responsible for checking the applicable deadlines and ensuring their submissions are completed on time.

08 | CONDUCT, QUESTIONS, AND REPORTS

• All participants must treat fellow students, organizers, and judges respectfully.
• Harassment, cheating, impersonation, fraudulent submissions, and deliberately misleading information are prohibited.
• Technical questions about the dataset, evaluation metric, and competition rules should be posted in the official Kaggle discussion forum whenever appropriate.
• Potential data leaks, security concerns, or methods of identifying anonymized samples must be reported privately to the Kaggle competition organizers. Such information must not be exploited or publicly disclosed.
• Questions about Devpost registration and research submissions may be directed to the hackathon organizers.
• The organizers may disqualify entries that violate the rules, subject to the applicable platform terms.

09 | OFFICIAL LINKS

Technical Competition and Detailed Rules:

OBSCURA on Kaggle

Hackathon Registration and Research Submissions:

OBSCURA on Devpost

Community and Discussion:

BTAJI Crew WhatsApp Community

After joining the community, participants can select the Kaggle group for OBSCURA announcements, technical discussions, and team formation.

By participating in OBSCURA, you agree to follow the official competition rules and applicable platform requirements.