Hockey Team Classification
Grouping/Sorting players into their respective teams
π Contribute : Found a typo? Something confusing? Have a suggestion? Please reach out
π΅οΈ Introduction
In this challenge you will be presented with over 2,000 cases of images from two teams participating in a youth hockey game. The goal is to separate those images into their respective teams.
While the solution can be simple or complex in nature, it should abide by the spirit of the challenge and be reasonably expected to perform just as well when additional teams are added to the dataset.
β± Timeline
- Start Time: 4:00 PM (UTC) 12th August 2020
- End Time: 4:00 PM (UTC) 30th September 2020
π Prizes
There will be 3 cash prizes;
- $2,500 USD
- $1,000 USD
- $500 USD
πΎ Dataset
The dataset provides 2,200 tests each include 10 images from two different teams. Each groups images are numbered from 1.jpg to 10.jpg and contain a randomized number of images from each team.
π Submission
- Prepare a csv container the following headers
- id ~ This can be extracted from the dataset path dataset/{id}/
- team_a ~ This is one of your groups, it doesn't matter which group you separate the teams into. This field should be a comma separated list of image indexes. The indexes can be extracted from the images names, `1.jpg` = `1`
- team_b ~ This is one of your groups, it doesn't matter which group you separate the teams into. This field should be a comma separated list of image indexes. The indexes can be extracted from the images names, `1.jpg` = `1`
- time (ms) ~ The time in milliseconds it took to run your code
π©βπ» Getting Started
To reduce boilerplate for getting started we have provided a repository that can help to get you started. The repository will handle setting up the dataset and a starting point for your code base.
π Evaluation Criteria
Primary Scoring
The challenge is made up of 2,200 groups of 10 images, in the primary scoring evaluation for every group that you properly sort all 10 images you get a point. The points are then divided by 2,200 to come up with a score between 0 and 1.
Secondary Scoring
The challenge is made up of 22,000 total images, in the secondary scoring evaluation for every image that you properly sort you get a point. The points earned are then divided by 22,000 to get a score between 0 and 1.
Tertiary Scoring
In the event a tertiary scoring is needed the average evaluation speed will be used to break a tie.
π Links
- πͺ Challenge Page
- π£οΈ Discussion Forum
- π Leaderboard
π± Contact
Participants
Leaderboard
01 | u1234x1234 | 0.852 |
02 |
|
0.482 |
03 | Dmitryakonovalov | 0.355 |
04 | DavidKazaryan | 0.273 |
05 | rdeggau | 0.188 |
Notebooks
0
|
0
|