Skip to main content

The first challange of Croatian AI league - dolphins recognition

Project description

Dolphins recognition challenge

"Croatian AI League"

Tutorial

The easiest way to start and submit your solution is to open the following tutorial on Colaboratory from Google. {% include note.html content='Click the button below to open a tutorial on Google Colaboratory. Clicking the button will open a Google Collaborative website with a tutorial containing detailed instructions and tips on how to train the model. By following the tutorial cell by cell, you will make your first model and participate in the challenge.' %}

"Open In Colab"

Intro

The challenge has two parts:

  • instance segmentation of dolphins in the photo (February 18th - June 4th 2021), and
  • recognition of an individual dolphin from the photo (June 5th - September 3th 2021).

Legal

  • Official rules (in Croatian) can be found at the following link.

  • Privacy notice (in Croatian) can be found at the following link.

Leaderboard

medal

The leaderboard will be periodically updated to reflect new sumbissions. {% include note.html content='The leaderboard is generated using validation set and will most likely be different when evaluated at the end of the contest using test dataset due to overfitting.' %}

    alias            date                          submitted_iou    calculated_iou
--  ---------------  --------------------------  ---------------  ----------------
 1  tekashi          2021-04-19 15:03:11.966772          0.51546               nan
 2  dolphinSantiago  2021-03-28 21:13:52.740719          0.48156               nan
 3  Orka             2021-04-02 16:21:59.548029          0.47683               nan
 4  alias            2021-05-02 10:38:49.415151          0.46234               nan
 5  Dupin            2021-03-31 14:35:22.171495          0.46228               nan
 6  dupincek         2021-03-29 17:04:06.266327          0.44912               nan
 7  Boto             2021-03-31 22:54:40.595595          0.44678               nan
 8  stokic           2021-02-21 18:51:53.232539          0.43552               nan
 9  dolphin_rovinj   2021-05-01 20:06:35.372099          0.43487               nan
10  firstML          2021-02-28 09:55:59.155951          0.42529               nan
11  rangoiv_0.0      2021-03-11 21:18:19.755089          0.41769               nan
12  testy            2021-03-12 15:06:14.210987          0.39001               nan

Local install

You can also work on your own personal computer or cloud by installing our pip package and downloading the data directly using the following command:

pip install dolphins-recognition-challenge

How to use

The dataset is prepared for use with PyTorch, although it is easy to prepare it for other frameworks. To use it with PyTorch, install the PIP package above and import it as follows:

from dolphins_recognition_challenge.datasets import get_dataset

train_ds, val_ds = get_dataset("segmentation")

The easiest way is to start is by downloading the same tutorial Jupyter notebook mentioned above.

Discussions

Discussions are hosted on Github, together with the complete source code of the challenge. There you can ask questions and discuss any matter you like.

Organizers

CISEx

CISEx is an association of independent software exporters that gathers over 280 members, mostly companies from IT sector oriented towards global market. Our main goal is to strengthen the position of IT sector in Croatia by supporting software production, export and entrepreneurship, creating new jobs and positioning our members on the global market. It serves as a platform for growth and development of all our members, joint appearance on the domestic and foreign markets, exchange of experiences and mutual cooperation.

Blue World Institute

The Blue World Institute of Marine Research and Conservation works to protect the marine environment in the Adriatic Sea. To that purpose, the Blue World Institute operates three programmes – research, education, and conservation. Our research focuses mainly on large marine vertebrates (dolphins and whales, sea turtles, sharks and giant devil rays) informing our education activities and conservation projects. We work from the Adriatic islands of Lošinj, Murter and Vis, with the local communities, and collaborate nationally, regionally and internationally to advance sustainable marine management and environmental sustainability in the Mediterranean Basin.

CroAI

CroAI, the Croatian Artificial Intelligence Association, was founded in 2019 and brings together leading companies and startups in the field of AI in Croatia. CroAI seeks to position Croatia as a country of unique opportunities for the development of human-centric AI through a culture of dialogue between entrepreneurs and decision makers at the national and European level.

Sponsors

Infobip

Infobip is a global cloud communications platform that enables businesses to build connected experiences across all stages of the customer journey. Accessed through a single platform, Infobip’s omnichannel engagement, identity, security and contact center solutions help clients and partners overcome the complexity of consumer communications, grow their business and increase loyalty. With over a decade of industry experience, Infobip has expanded to include 65+ offices on six continents offering natively built technology with the capacity to reach over seven billion mobile devices and ‘things’ in 190+ countries connected directly to over 650 telecom networks.

Photomath

Photomath is a fast-growing EdTech company whose mobile app is the #1 app in the world to learn math. Powered by advanced machine learning technology, app instantly scans, accurately solves, and intuitively explains printed and handwritten math problems to users through step-by-step explanations. With over 220 million downloads globally and tens of millions of users worldwide every month, Photomath is the most popular mobile application from Croatia and one of the most popular educational apps of all times. The Photomath team in Zagreb has grown to over 100 employees and more than 100 students working on content creation, and they also have a team of a dozen people in Silicon Valley, USA. Photomath’s mission is to democratize education through the use of artificial intelligence and software-generated educational content. For their ambitious plans, they bring together top experts who have the opportunity to work on a global product and gain unique technological knowledge and experience. In 2021, they plan further investments and employment, especially in the segment of software development and artificial intelligence.

Porsche Digital Croatia

Porsche Digital Croatia is a joint venture formed by Infinum, the leading software design and development company in Croatia, and Porsche Digital, a subsidiary of the German sports car manufacturer. The goal of this company is to be the hub of digital innovation and high-tech solutions in the automotive sector, and the main mission is to find and develop new digital business models and optimize existing products. Their teams of talented professionals are focused on technological innovation and are passionate about software development and design, machine learning and artificial intelligence.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

dolphins_recognition_challenge-0.0.16.tar.gz (32.0 kB view details)

Uploaded Source

Built Distribution

File details

Details for the file dolphins_recognition_challenge-0.0.16.tar.gz.

File metadata

  • Download URL: dolphins_recognition_challenge-0.0.16.tar.gz
  • Upload date:
  • Size: 32.0 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/3.4.1 importlib_metadata/3.10.0 pkginfo/1.7.0 requests/2.25.1 requests-toolbelt/0.9.1 tqdm/4.59.0 CPython/3.8.8

File hashes

Hashes for dolphins_recognition_challenge-0.0.16.tar.gz
Algorithm Hash digest
SHA256 9766c6f776b3b2a5a664cc5f14884b25834e75c8998fd754205d5f31850a5bd0
MD5 484bc2126d4110de77030b80e1e311f4
BLAKE2b-256 b72f6b33cec968a74a61bbc708c170fa965c7ded7c558856bf091baaa4aa1df9

See more details on using hashes here.

File details

Details for the file dolphins_recognition_challenge-0.0.16-py3-none-any.whl.

File metadata

  • Download URL: dolphins_recognition_challenge-0.0.16-py3-none-any.whl
  • Upload date:
  • Size: 31.2 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/3.4.1 importlib_metadata/3.10.0 pkginfo/1.7.0 requests/2.25.1 requests-toolbelt/0.9.1 tqdm/4.59.0 CPython/3.8.8

File hashes

Hashes for dolphins_recognition_challenge-0.0.16-py3-none-any.whl
Algorithm Hash digest
SHA256 bd87286f26cedf138674279e864a55c7cb38ef71541503fcdcf1e1bc2e1360ba
MD5 a6bd54309e45f8aedd68d55fa13ee2ee
BLAKE2b-256 f43fc86cab0e0ae415519ea75c281ab62bf90fbc137b3edd3c37bb1d9e0b1eb3

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page