# Architectural Drawings Recognition and Generation through Machine Learning (2018)
URL: https://www.associativetrails.com/writing/2022/05/01/architectural-drawings-recognition-and-generation-through-machine-learning-2018/
Published: 2022-05-01 · Practice · Sheet 04
Summary: Notes on a 2018 paper by Weixin Huang and Hao Zheng

The paper Huang & Zheng · 2018

ACADIA (http://papers.cumincad.org/cgi-bin/works/paper/acadia18_156) ResearchGate (https://www.researchgate.net/publication/328280126_Architectural_Drawings_Recognition_and_Generation_through_Machine_Learning) PDF 6.5 MB (https://www.associativetrails.com/assets/uploads/05.acadia18_156.pdf)

- Pix2PixHD (https://github.com/NVIDIA/pix2pixHD) used to generate zoning maps of floor plans
- Detail improves with number of epochs - ResNet used to prevent overfitting
- CGAN - Conditional Generative Adversarial Networks - punish trining going in wrong direction
- Pix2PixHD - 5 groups of convolutional layers, 9 groups of residual network layers, 5 groups of deconvolutional layers
- Using PyTorch - images from lianjia.com (https://bj.lianjia.com/)
- Labels used: Walkway, Bedroom, Living Room, Kitchen, Toilet, Dining Room, Balcony, Window, Door
- 115 image pairs - 100 training, 15 testing ~ 2hrs to train one network
- Works well in recognising bedroom, kitchen, toilet & balcony - normally well-defined by walls and specific furniture
- Less successful for walkway/living room - but some architects got this wrong when labelling test data (done by Architectural students)
- Struggled with non-orthogonal shapes
- In one image, GAN found mis-labelled data. Later found 4 errors in human-labelled data, which didn't lead to training in the wrong direction - pretty robust network.
- Generating plans from zones - sometimes put furniture facing the wrong way. Some architects agreed with the network's furniture position above the position in the test images.
- 2-pixel convolutions. Residual network avoids overfitting.
- Trained to 80 epochs - loss value stabilised after that
- The paper looks at individual neurons' feature detection
- Bedrooms/balconies often activated together