SpikeCV-WUJI

📅 Competition Timeline

📢 April 18, 2025: Competition announcement & Registration
📝 May 1, 2025: Submission begins (Track 1 starts ranking)
⏰ July 10, 2025: Deadline of Main, Dataset & Innovation tracks. Submit demo video + code/data
🏆 July 25, 2025: Results announced
🎉 August 28–31, 2025: Award Ceremony @ IJCAI Satellite Event, Guangzhou

⏳ Starts in

⚠️ Notice (Fairness Statement)

To ensure fairness, participants from PKU-related institutes who have published relevant work since 2020, or are involved in spike vision research at these institutes, are not eligible for awards.
Publications based on competition data or models must acknowledge SpikeCV and cite the competition reference.

💰 $15,000 Prize Allocation

🎨

Texture Restoration

🥇 $4,000

🥈 $2,500 (x2)

🥉 $1,500 (x3)

💾

Dataset & Benchmark

🏅 Best Award: $2,000

💡

Open-Spike Innovation

💡 Best Award: $1,500

🌍

Outstanding Participants

🙋 $300 (x5)

📈 Scoring Rules

The scoring of each track is independent.
Track 1 will be evaluated by a mix of automated metrics, Track 2&3 will be evaluated by expert reviews.
Furthermore, Outstanding Participants will be awarded based on Track 2&3. For any questions regarding the rules, feel free to contact us at spikecv@outlook.com.

Track 1: Texture Restoration

Track 1 includes four tasks: Hybrid Camera Deblur, Deocclusion, Color Reconstruction and 3D Reconstruction.
For each task, multiple metrics and their weights will be announced, including reference-based and no-reference-based metrics.
Participants' score of each task will be based on the rank they get on the corresponding task, as listed below:

Rank	1st	2nd	3rd	4th	5th	6th	7th+
Score	23	11	5	2	1	1	0

Participants's final score in Track 1 is the sum of the points they get on each task.
If multiple teams get the same final score in Track 1, they will equally share the total amount of the tied prizes, and the team with fewer participants will be ranked higher on the leaderboard.
For example, a team challenges three tasks on Track 1, and get 2nd on Hybrid Camera Deblur, 3rd on Deocclusion, 5th on Color Reconstruction, their final score is 11+5+1=17.

Track 2: Dataset & Benchmark

A weighted ranking rule based on expert review scores for dataset contribution.
Datasets demonstrating strong generalization on real-world data are highly recommended.

Track 3: Open-Spike Research

A weighted ranking rule based on expert scores for research innovation.
Proposals that inspire spike vision to explore new applications and stimulate future tasks are highly recommended.