Embedded Files
2nd AVA Challenge launched @ IEEE MIPR 2024!
News
News
The competition will start on May 20!
AVA Challenge: Anticipating Vehicle Accidents Challenge.
AVA Challenge: Anticipating Vehicle Accidents Challenge.
Challenge Description
Challenge Description
Research in video analysis plays a vital role in autonomous driving by aiding in accident prediction. Dashcam footage provides valuable insights for this purpose. However, there's a lack of well-annotated, extensive datasets for accident prediction in Asian regions.
To address this issue, we collected a large-scale dataset of vehicle collisions from Taiwan and annotated it. The objective of this challenge is to predict the risk of an impending car accident to the recording vehicle.
Context
Context
According to the research, self-driving cars can only reduce about one-third of major car accidents that occur in the United States out of 5,000 incidents. People typically assume that autonomous driving systems can eliminate human errors while driving. However, many car accidents are still caused by human factors, including perception errors, driver distraction, poor visibility, or slow reactions. Other possible causes are prediction errors, such as misjudging the distance to other vehicles or pedestrians' movements. Some errors may be related to planning and decision-making, such as inappropriate evasive actions or overcompensation when controlling the vehicle. Other accidents can be caused by impairment due to alcohol, drugs, or drowsiness. There are avoidable accidents that are caused by vehicle malfunctions.
Based on the above, we propose a topic for participants to utilize existing data to predict the risk of an impending car accident to the recording vehicle, in order to assist in crisis response for autonomous driving and enhance traffic safety.
The competition host is Advanced Computer Vision Lab (ACVLab), a research team led by Professor Chih-Chung Hsu at National Cheng Kung University. Our lab focuses on introducing machine learning/deep learning into various applications in computer vision and cultivating students with high practical and applied capabilities in deep learning.
Your efforts in this competition could help extend the benefits of road safety. Greater access could further reduce the occurrence of car accidents worldwide.
Challenge Website Link
Challenge Website Link
Kaggle is our competition platform. Please refer to Kaggle: https://www.kaggle.com/competitions/2nd-ava-challenge-ieee-mipr-2024.
Timeline
Timeline
May 20, 2024 - Start Date.
June 30, 2024 - Final Submission Deadline .
July 1, 2024 - Private Leader Board Announce.
July 7, 2024 - Technical Report Submission, Completed training code and related materials deadline.
July 14, 2024 - Competition End Date - Winner's announcement.
All deadlines are at 11:59 PM UTC+8 on the corresponding day unless otherwise noted. The competition organizers reserve the right to update the contest timeline if they deem it necessary.
Evaluate
Evaluate
Submissions are evaluated on area under the ROC curve(roc_auc_score) between the predicted probability and the observed target.
Submit format
Submit format
For each "file_name" in the test set, you must fill in the "risk" column with predictions (float).
You may refer to the dataset for an example submission file.
Rules
Rules
One account per participant
One account per participant
You cannot sign up to Kaggle from multiple accounts and therefore you cannot submit from multiple accounts.
No private sharing outside teams
No private sharing outside teams
Privately sharing code or data outside of teams is not permitted. It's okay to share code if made available to all participants on the forums.
Team Mergers
Team Mergers
Team mergers are allowed and can be performed by the team leader. In order to merge, the combined team must have a total submission count less than or equal to the maximum allowed as of the Team Merger Deadline. The maximum allowed is the number of submissions per day multiplied by the number of days the competition has been running.
Team Limits
Team Limits
The maximum team size is 5.
Submission Limits
Submission Limits
You may submit a maximum of 5 entries per day.
You may select up to 2 final submissions for judging.
External Materials
External Materials
The use of external materials that are publicly and freely available for research or academic purposes is permitted, but the source must be clearly stated in the technical report.
Awards & Recognition
Awards & Recognition
After the competition concludes, we will invite the top 6 participants to submit technical reports/code for verification. Upon successful verification, we will then invite the three participants to co-work on this challenge summary paper, as well as to present (or virtually) their solutions at the IEEE MIPR2024 conference.
* The technical report requires the MIPR paper format template, with a minimum of 2 pages.
Our sponsor, PHISON, is providing the following prizes:
1st place: SSDs valued at approximately $500
2nd place: SSDs valued at approximately $300
3rd place: SSDs valued at approximately $200
Organizers
Organizers
Prof. Chih-Chung Hsu - Institute of Data Science, National Cheng Kung University, Tainan, Taiwan
Prof. Ming-Ching Chang - Department of Computer Science, University at Albany, State University of New York, USA
Sponsers
Sponsers
Page updated
Google Sites
Report abuse