A selection of thoughts, notes on other people's thoughts, experiences with APIs, tips, tricks, and other stuff.
Using a convolutional neural network to recognise top-level architects' individual styles. Almost 70% or architects can be distinguished with 80% accuracy. 20,000 sample images by 34 past Prisker Prize winners.Read more
A generative adversarial network to superimpose coloured rectangles on architectural floor plans based on room type (Bedroom, kitchen, etc.). Worked really well for some rooms - it even picked up a few mistakes in the human labelling.Read more
An attempt to build a convolutional neural network to classify 1000 images into 3 categories. Accuracy peaked at 82%, no hyperparameter tuning was performed.Read more
Development of a convolutional neural network for the recognition and labelling of rooms within floor plan images of multi-residential apartments. Trains on 454 floor plan images sorting into 6 categories - claims the results were "satisfactory".Read more
A comparison of different machine learning methods and their accuracy in categorising building use. The deep learning models came out on top, with the model based on ResNet50 performing particularly well. 240 images were used to train the models to sort into three categories.Read more