Decoded Disorder// S.3 SOFTWARE


The idea of this project is upcycle planarised wood products such as palettes, offcuts and packaging wood forming the majority of this waste. Using supervised machine learning to identify and classificate wood planks. The input images will be analyzed by computer vision by using OpenCV to get dimensions of meterial and a series of features which will feed into neural network. The algorithm will classficate woods into three grades as good, medium and bad.




In this system a QR code scaning python script is developed to get the croped scanned planks image then process the images in three different ways. All these values are collected in the database for desgin.


In order to get the dimensions of the plank, we placed a QR code on the top, which can helps extract the pixel size rate for each image then calculate the dimensions by math equation.


Feature Detection

After cropping the image, we break down the machine learning parameters into these four visable valuables which are nail holes, short cracks, long cracks nad knots. The algorithm categorize the objects in terms of their size and shape.


Usable Length 

Due to the variation of defects of recycled wood and recycle them as much material as possible, we developed these algorithm to divide the plank into segments if the segment wide has more than % deviation from maximum length, it will recongnized as unusable part.

Machine Learning

After all the computer vision processing, we print the values in a CSV sheet which will be read by OWL in grasshopper. In this machine learning step, we construct three onehot tensors according to our grading system, the output will be the confidence of each grade. Here we test 50 planks and the graph below shows the result of prediction.


Decoded Disorder // S.3 Software  is a project of IAAC, Institute for Advanced Architecture of Catalonia developed at the Master in Robotics and Advanced Construction in 2022 by:

Students: Huanyu Li, Alfred Bowles, Shamanth Thenkan, Vincent Verster

Faculty: Mateusz Zwierzycki
Faculty Assistant: Nikol Kirova