Methods for automated digital reading of colorimetric sensors in settings with perturbed illumination conditions and low image-resolutions.
The proposed methodology aims to improve the accuracy of colorimetric sensor reading in altered illumination conditions.
We implement image processing and deep-learning (DL) methods that correct for non-uniform illumination alterations and accurately read the target variable from the color response of the sensor.
── 1.1 Dataset Generation
── 1.2 Dataset Noise Augmentations
── 2.1 Model Zoo
── 2.2 Image Generator Zoo
── 2.3 Train Multi-task autoencoder with latent regression
── 2.4 Train MLP for color-based prediction
- Clone the repository
git clone 'git@github.com:SchubertLab/colorimetric_sensor_reading.git'
- Install dependencies
conda env create -f environment.yml