Welcome to the Whatever Hack hackathon March 16

In our Moscow office on March 16, a 48-hour machine learning marathon will begin, during which you can demonstrate your skills to create a solution for generating memes, distinguish a wardrobe from a carpet, or make the best MVP (minimum viable product) within a free theme.

We invite everyone to solve any of the three tasks and fight for a prize fund of 1.5 million rubles (or not rubles).

Further - about the tasks and details.

The rules are very simple. Each team can have up to 4 people, with the in-person presence of all team members. Participants are invited to decide on the choice of one of three problems:

1. Determining the quality of memes

Dataset is a list of pictures and text captions for them. For each pair of picture – caption there is a label - shared with it or not. It is required to predict whether people will share a particular image-signature pair or not.

2. Detection and classification of furniture in interior photos

Dataset consists of photographs in which items of furniture are marked with a bounding box. It is necessary to do ~ git clone faster-rcnn ~ , to train the detector, which will find the given pieces of furniture, and in the future will help to choose apartments by parameters (to make a filter “without grandma's carpets”).

3. Free theme

It will be possible to solve problems that have been postponed long ago, or to solve the expanded version 1 and 2 tasks.

The first two tasks will be an objective metric. In the third task, the decisions of the participants will be assessed by the jury.

The five best teams in each category are awarded prizes:

1. 240k ₽
2. 120k ₽
3. 80k ₽
4. 40k ₽
5. 20k ₽

However, the prize can be obtained in the crypt.

Hackaton Whatever Hack is timed to the birthday of the largest ML community in the CIS - Open Data Science - and is carried out with the support of the Dbrain and Mail.Ru Group. Participation is free, but you need to register .

Collection: Moscow, March 16, at 18:00, Mail.Ru Group office (Leningradsky prospect, 39, p. 79).

Completion: March 18.