This month Meetup was entirely devoted to the students who have implemented machine learning algorithms for developing interesting AI solutions during Hackathons (a weekend-long intense competition for programming). At such hack marathons, most of the participants are developing their applications and prototypes without much sleep but with much passion to turn their idea into reality.
This is what the students from TUES managed to do – turning data into meaningful insights. For those of you who couldn’t make it to the venue, you can still catch up with:
- Developing an app implementing sentiment analysis and topic modeling (Team Peper’s presentation + GitHub)
- Object detection model with YOLO and DarkFlow ( Team Nishki’s presentation)
- MeetUp Video in YouTube
About the teams:
Team “Peper” ( Victor Velev, Vladislav Georgiev, Kaloyan Madzhunov, Martin Dacev, Peter Milev, and Telerik Arsov – 12th and 10th grade students at Technological School “Electronic Systems” (TUES) at the Technical University of Sofia) has implemented the “Peper Analysis” application that makes a quick and user-friendly analysis of the neighborhoods in Sofia in the form of a heat map for tourists to find a suitable place to stay in Sofia by searching for different subjective criteria. The solution is implemented through web scraping (Selenium), sentiment analysis (TextBlob) of the acquired data, analysis (NLTK), topic modeling and word embedding (GloVe).
Team Nishki (Evgeni Dimov, Kalin Doichev, Kostadin Kostadinov and Aneta Tsvetkova from TUES) were solving a case about identifying objects with their labels through image recognition algorithms such as YOLO for real-time object detection and classification while the implementation in DarkFlow. The team was among the finalists of the international Data Science hackathon happening in 2019.