Opencampus.sh Machine Learning Projects

The projects people are building with the machine learning courses at opencampus.sh

39 projects

A Hybrid Approach to Predict Cryptocurrency Value

Machine Learning with Tensorflow, SoSe 2021The goal of this project is to train a model which is able to classify if the value of specific cryptocurrency would rise or decline in the next day and would it be rational to buy or sell that crypto.Anita Safi, Rakeshwer Rakeshwer, Awais Khan Malik

Age from Face

Deep Learning from Scratch, WiSe 2122Predict age of a person from an image of their faceOskar Lachnit, Florian Diedrich, Nils-Christian Buck

Automated Essay Scoring

Machine Learning with Tensorflow, SoSe 2021Automated scoring of social study evaluationSabrina Ludwig, Christopher Hansen, Christian Mayer, Kerstin Eilers

Bakery Sales Prediction (G1)

Deep Learning from Scratch, SoSe 2021Deep Learning project to predict future sales of a bakery store according to products.Johannes Dupont, Pavan Jethvani, Osama Alrabayah, Jannik Knopp

Bakery Sales Prediction (G2)

Deep Learning from Scratch, SoSe 2021Prediction of bakery sales based on Metolytix DatasetAmelie Schmuecker, Samira Ferssioui, Niko Schmidt, Christopher Hansen

Bike Sharing Prediction (SprottenFlotte)

Deep Learning from Scratch, WiSe 20/21Analyzing the SprottenFlotte data from Kiel, can we predict where will the next bike be borrowed? Note: the data for this project is private and cannot be shared, only the results.Andrej Ponomarenko, Daniel Michells

Chess game rating quantification

Advanced Deep Learning - from CNN to Transformers, SoSe 2021The idea is to predict, who wins a chess game after some moves. That could be a first step to analyse, which move(s) where the main reason for losing a game or not winning it.Niko Schmidt

Classification of illustrations in historic monographies

Machine Learning with Tensorflow, WiSe 20/21Digitalizing old collections makes them available to a worldwide public. This project trains a CNN to identify illustrations in monographies from the 15th to the 18th century.Irena Kampa

Climate Change

Deep Learning from Scratch, SoSe 2021We try to simulate Climate Scenario Analysis using 3 different IPCC climate scenarios; RCP3PD, RCP4.5, and RCP6 and predict temperature for each scenario in the year 2100.Vineet Shah, Shilpika Sarvepalli

COVID-19 Detection on X-Ray Image

Deep Learning from Scratch, WiSe 20/21Can we detect COVID from an X-Ray image of the lungs? It turns out we can, or better, a software can do that for us. This project achieved an accuracy of 0.9896 on this task!Mithun Das, Mohammad Wasif Islam, Rakibuzzaman Mahmud, Sarker Miraz Mahfuz

Designing Airfoils

Machine Learning with Tensorflow, WiSe 20/21Airfoils usually are designed using complex differential equations. Can we feed a NN with a set of 2D outline of airfoil profiles and let it learn?Nils Berns, Violetta Germann

Disease Classification on Medical XRay Images

Deep Learning from Scratch, WiSe 20/21Can a neural network distinguish different types of diseases just by looking at a single X-Ray image of the lungs? The project shows promising results in this direction.Sudesh Acharya

Face Mask Recognition

Deep Learning from Scratch, WiSe 20/21This project develops a detection system which tells from a picture whether the person is wearing a mask or not.Adnan Nooruddin, Ravish Kumar, Christoph Eberz, Bennet Möller

Fake News Detection

Deep Learning from Scratch, SoSe 2021This project uses a Fake and real news dataset from the Kaggle. We implement a Feed Forward Neural Network and a LSTM techniques using TensorFlow to find the desired output for the project.Farjana Islam, Junaid Abbas Mohammed, Md Majharul Islam Nayem

Fake News Detection Onion vs Nottheonion

Deep Learning from Scratch, WiSe 2122The project contains a model, which learned to predict whether news headlines are either real or made up and fake.Peyman Kazemi, Max Kaapke, Alexander Suhm

Fine-tuning GPT-2

Natural Language Processing, WiSe 20/21Can GPT-2 be fine-tuned to achieve GPT-3 like accuracies? The project fine-tuned different models and compared them on different tasks.Steffen Brandt, Steffen Pohle, Atul Kumar Yadav, Jonas Peltner, Nicolas Steen, Philipp Seeler

Flower Classification

Deep Learning from Scratch, WiSe 2122The main purpose of this project is flower classification using transfer learning. The study of flower classification is an important aspect due to its complex environment for extracting features, similarity of features in many species, changing features in the life of a flower.John Jay Kimani

Generating Fake Chinese Characters by GAN

Generative Adversarial Networks, SoSe 2021The goal of this project was generating images of artificial / "fake" Chinese Characters by using a generative adversarial network (GAN) trained on a dataset of images containing real Chinese characters.Anita Safi, Sebastian Koch

GPT-NEO United Nations

Machine Learning with Tensorflow, SoSe 2021The goal of this project is to finetune a generative pretrained transformer (GPT) and generate United Nations (UN) resolutions.Felix Gröner, Nils Axe

Hydrochemistry Time Series

Deep Learning from Scratch, SoSe 2021We try to predict the amount of chlorophyll-a using the Boknis Eck timeseries dataset. This dataset comprises a number of measurements performed in Boknis Eck, close to Eckernförde, over the course of 50 years. We use a three layers dense neural network to predict our parameter.Julia Börth, Matthias Bressan, Irena Kampa, Wanja Böhme

Iceberg and Ship Detection in Radar Satellite Imagery

Machine Learning with Tensorflow, WiSe 20/21This project is aimed at building an algorithm able to detector for SAR imagery which finds and classifies ships, icebergs or unidentified objects.Eike Schütt, Yi-Jie

Image classification for birds

Machine Learning with Tensorflow, SoSe 2021This project aimed to configure and train a network to predict the specie of a bird based on its image using different computer vision training techniques including CNN and transfer learning.Esam Al Sakka, Kai Paulsen

Image Segmentation

Deep Learning from Scratch, WiSe 20/21Segmentation is used to separate an object from the background. Using a U-Net the group was able to do some interesting work on a Kaggle Challenge.Yi-Jie Yang, Sebastian Koch, Erwin Smith, Suman Singha

Music genre classification task

Machine Learning with Tensorflow, WiSe 20/21Have ever had a discussion with your friends about the genre of a song? Here the solution: a neural network to automatically classify musical genre!Rakibuzzaman Mahmud, Sarker Miraz Mahfuz, Mohammad Wasif Islam

Network traffic prediction

Machine Learning with Tensorflow, WiSe 20/21Internet traffic accounts for huge quantities of data, yet sometimes it's hard to make sense of them. This project aims at classifying network traffic using passive measurements of TCP connections.Mithra Gholami

Painting Classification

Deep Learning from Scratch, WiSe 20/21Which artist painted this painting? Nils and John tried to answer this question using neural networks with different approaches, discover more in their presentation.John Jay Kimani, Nils Berns

Phishing Website Detection

Deep Learning from Scratch, SoSe 2021Using Natural Language Processing approach and by focusing on words in the address Bar, a Neural Network has been traind to detect Phishing and legitimate website.Anita Safi, Manikwadura Dilini Tharindika, Awais Khan Malik, Razeeb Sarker

PM10_Prediction

Deep Learning from Scratch, WiSe 2122A ML project to predict PM10 pollution at selected air monitoring stations. The machine should be trained with weather data and PM10 pollution data from the past. Using weather forecast data, the machine should predict PM10 pollution.Anke Schürmann, Philipp Jordt, Friedrich Kerchnawe

Predicting Bakery Turnover using AR-Net

Machine Learning with Tensorflow, WiSe 20/21How accurately can we predict the bakery turnover? This group got an auto-regressive neural network on this task!Ravish Kumar, Modeus Abdelnaby

Real-Time Object Detection with YOLO[v3/v4]

Advanced Deep Learning - from CNN to Transformers, SoSe 2021Object Detection with YOLO[v3/v4], using OpenCV and PyTorch, undertaken for the partial fulfillment of Advanced Deep Learning Module ( SS2021 ) and ETCS Creditpoints @OpenCampus.sh.Sudesh Acharya, Bennet Möller

Shopping Receipt Analysis

Deep Learning from Scratch, WiSe 2122We tried to detect and read the total sum from a picture of a shopping receipt. Based on a dataset of self taken pictures of shopping receipts.Note: data has been redacted for privacy purposes.Valentin Buck, Jannik Knopp, Irena Kampa

Single Neuron Signal Analysis

Deep Learning from Scratch, WiSe 2122This project aims at automatically classifying signals recorded from neurons in the human brain. We implemented two neural networks to classify spike sgnals.Note: Data was provided by Marcel Kehl and Florian Mormann from the Department of Epileptology, University of Bonn Medical Center.Anton Krieger, Katsiaryna Tsishchankava

stock market news analysis by NLP

Machine Learning with Tensorflow, WiSe 20/21Are daily news correlated with the stock market? Can we use them to predict the market behaviour? Here some interesting attempts!Andrej Ponomarenko, Monfrared Gharibi Foorogh

Study Recommender System with GPT-3

Natural Language Processing, WiSe 20/21How can we choose the best study program for us when there are so many choices? Jan, Jan and Erwin built a system using state of the art technology to recommend you the best choice based on what you are looking for.Jan Peter Prigge, Jan Deller, Erwin Smith

Surf Spot Prediction (Windfinder)

Deep Learning from Scratch, SoSe 2021A basic machine learning approch on predicting the pageviews for the windfinder website when provided with weatherdata. Note: the data for this project is private and cannot be shared, only the results.Kilian Köln

Surf Spots Predictions (Windfinder)

Deep Learning from Scratch, WiSe 20/21Predicting the best spot for surfing is a hard task, yet Lennart gave it a great shot and developed a model to predict it with 86% of accuracy. Note: the data for this project is private and cannot be shared, only the results.Lennart Petersen

Using containership positions (AIS) to predict export statistics

Machine Learning with Tensorflow, WiSe 20/21The project worked on training a nueral network for the prediction of unilateral export statistics using Neuralnet and AIS-Containership dataJakob Stender, Steffen Gans

Voice Detection (Classify Human based on Voice)

Machine Learning with Tensorflow, WiSe 20/21This group focused on voice detection and human classification based on the voice.Mutasim Fuad, Mithun Das, Rajib Chandra Das

Writing System Recognition

Machine Learning with Tensorflow, WiSe 20/21Can we detect different language from an image? Here a classifier which is able to distinguish Latin, Chines, Kyrillic and Georgian!Manpreet Singh, Adnan Nooruddin, Rahima Akter, Sebastian Koch