machine learning model

The smart bin is able to recognise and classify trash into different categories.
When people sort their trash correctly, the installation - made of plastic objects, will be activated as an alive and joyful helper, moving and creating sound.
On the other hand, if individuals don’t recycle properly, the creature will be transformed into a monster.
Multi Label Classification (MLC) model developed by the machine learning enginner Israel Goytom.
The MLC model is used to classify the material of the waste as plastic or other types. Using a pre-trained model which is trained on ImageNet with 1.2 million images as a feature extractor and then a dataset of plastic waste consists of 1000 images (open source), we trained a deep learning model for classification and detection. The model was trained on NVIDIA V1000 GPU for 6 hrs and it has 91% accuracy.

Animation of the interaction of people with the bin and the reaction of the installation and the sound.