Jacaranda Caroba - Hyperspectral Confocal Laser Scanning Microscopy (CLSM) dataset
Dataset with 148 hyperspectral images captured with resolution 2048x2048x32, and pixel precision of 16-bits, by CLSM with a Zeiss LSM 780. There is also a RGB version with dimension 2048x2048x3 and 8-bit pixels. There are 6 classes: Control, 5mg/L, 15mg/L, 30mg/L, 60mg/L and 1162mg/L (check file names). They correspond to Jacaranda Caroba leaf cuts of specimens exposed to potassium fluoride (KF) spray for a period of 60 days.
Neural Networks <-> Complex Networks dataset
Data and source code related to the work "Structure and Performance of Fully-Connected Neural Networks Through Complex Networks", which introduces the concept of Bag-Of-Neurons (BON) for analyzing neurons on fully-connected neural networks using Complex Networks. Reference: Scabini, Leonardo FS, and Odemir M. Bruno. "Structure and Performance of Fully Connected Neural Networks: Emerging Complex Network Properties." arXiv preprint arXiv:2107.14062 (2021).
KTH-TIPS 2 subset (MNIST-like)
The original KTH-TIPS 2 dataset comprises RGB images of 10 texture classes: aluminum foil, cork, wool, corduroy, linen, cotton, brown bread, white bread, wood, and cracker. We built a different dataset by cropping the original samples with size 200x200 (other sizes are not considered) into 28x28 non-overlapping gray-scale images. We divide the new crops into training and testing folds by ensuring that no same image crops appear together in the same fold. This yields a total of 165 thousand samples for training and 35 thousand for test; thus, we randomly choose 60 thousand images for training and 10 thousand for test. Reference: Scabini, Leonardo FS, and Odemir M. Bruno. "Structure and Performance of Fully Connected Neural Networks: Emerging Complex Network Properties." arXiv preprint arXiv:2107.14062 (2021).
download: KTHTIPS_balanced_grey_28x28.pickle
Detection of a SARS-CoV-2 sequence with genosensors using data analysis based on information visualization and machine learning techniques
Download the original SEM images and the descriptors computed from them using various Computer Vision techniques (AHP, CLBP, CNTD, Fourier, Fractal, GLDM, LCFNN, MobileNet, DenseNet201, InceptionResNetV2)
Machine learning and image processing to monitor strain and tensile forces with mechanochromic sensors
The static images are present in this repositoy (folder "cropped_images"), as well as the obtained segmentation masks (folder "new_masks"). The segmentation process is implemented on “pre_processing.py”. We also make the source code of the machine learning procedure available at: ‘machine_learning.py”. The folder "videos" contains the videos used in the paper. Reference: LDC de Castro, L Scabini, LC Ribas, OM Bruno, ON Oliveira Jr. Machine learning and image processing to monitor strain and tensile forces with mechanochromic sensors. Expert Systems with Applications 212, 118792