Patents and Publications
- O. Boiko, A. Collard, C. RENÉ, E. SOLEIMANI
Image processing methods and systems for low-light image enhancement using machine learning models
This work focuses on an image processing method designed to enhance illumination in input images representing various scenes. The method involves down-sampling the input image, processing the down-sampled image using a pre-trained machine learning model to produce a multiplicative correction map, and applying this map to enhance illumination. The correction map contains multiplicative factors tailored to improve the down-sampled image’s brightness. After up-sampling the correction map, an output image is generated by multiplying the original input image with the up-sampled correction map, resulting in enhanced illumination.
- O. Boiko, A. Collard, C. RENÉ, E. SOLEIMANI
Image processing methods and systems for generating a training dataset for low-light image enhancement using machine learning models
The objective of this work is to develop an image processing method for generating training datasets aimed at enhancing the illumination of input images using advanced machine learning techniques. This involves creating target image and low-light image pairs to train the model effectively. The approach enables the generation of high-quality training data that supports the improvement of image illumination, facilitating robust model performance in low-light conditions.
- O. Boiko, J. Hyttinen, P. Fält, H. Jäsberg, A. Mirhashemi, A. Kullaa, and M. Hauta-Kasari
Deep Learning for Dental Hyperspectral Image Analysis
27th Color and Imaging Conference Final Program and Proceedings (CIC27), pp. 295-299(5)
The aim of this work is automatic and efficient detection of medically-relevant features from oral and dental hyperspectral images by applying up-to-date deep learning convolutional neural network techniques. This will help dentists to identify and classify unhealthy areas automatically and to prevent the progression of diseases. Hyperspectral imaging approach allows one to do so without exposing the patient to ionizing X-ray radiation. Spectral imaging provides information in the visible and near-infrared wavelength ranges.
The dataset used in this paper contains 116 hyperspectral images from 18 patients taken from different viewing angles. Image annotation (ground truth) includes 38 classes in six different sub-groups assessed by dental experts. Mask region-based convolutional neural network (Mask R-CNN) is used as a deep learning model, for instance segmentation of areas. Preliminary results show high potential and accuracy for classification and segmentation of different classes. - To be continued … :)