This repository contains 286 raw original images of the raspberry trays, each with a resolution of 3948 X 2748 pixels. Each raspberry tray can have two diseases, such as albinism and fungus-rust, or two defects like over-ripeness and peduncle. Also, it contains a group of preprocessed images with bounding boxes that contain the ground truth about each annotated disease or defect in the fruit: 786 albinism labels, 164 fungus-rust labels, 164 fungus-rust labels, and 244 peduncle labels. Additionally, the MATLAB code is available to apply three approaches: descriptive statistics, statistical modeling, and convolutional neural networks.

Synthetic-sPVDB is the largest dataset of palm vein images of the state-of-the-art, comprising of 20,000 subjects with 6 samples per each. The synthetic images were generated by using the StyleGAN model and all of them were compared against all images in the synthetic dataset aiming to ensure that each individual is unique in the database. Finally, the main samples were increased by applying sample augmentation. Since palm vein recognition systems only use a region-of-interest (ROI) from the image of a human hand, we concentrate our study on generating ROI samples of palm vein images. Images are labeled as follows: ID_sample.png where "ID" is a 5-digits number representing an individual’s ID, and “sample” stands for the sample number from 1 to 6 (6 samples per individual). The format of the images is PNG, and the size is 128x128 pixels.

To request a copy of the database, the investigator must fill out the License Request Form (Google Form). Sign the License Agreement (https://bit.ly/3rubJxv), and upload it in the form's last question. Once the request has been processed, a download link will be provided by email. Please, notice that students are not eligible; therefore, the Form must be completed and signed by your supervisor.

NS-PVDB comprises 16,000 subjects with six samples each, which were obtained by applying sample augmentation. Since palm vein recognition systems only use a region-of-interest (ROI) from the image of a human hand, we concentrate our study on generating ROI samples of palm vein images. The synthetic images were generated based on a natural transport network to simulate the palmar vascular network and generate palm vein images for biometric recognition purposes. All generated images were compared against all samples in the synthetic dataset to ensure that each individual was unique in the database. Images are labeled as follows: ID_sample.png where "ID" is a 5-digits number representing individual’s ID, and “sample” stands for the sample number from 1 to 6 (6 samples per individual). The format of the images is PNG, and the size is 128x128 pixels.

To request a copy of the database, the investigator must fill out the License Request Form (Google Form). Sign the License Agreement (https://bit.ly/3t34r4r), and upload it in the form's last question. Once the request has been processed, a download link will be provided by email. Please, notice that students are not eligible; therefore, the Form must be completed and signed by your supervisor.

In order to stimulate research in writer verification and recognition by using handwritten strokes, the Laboratory of Technological Research in Pattern Recognition of the Universidad Católica del Maule (LITRP) makes the LITRP-SGDB database publicly available to the scientific community.

This repository contains images of 5 types of simple graphemes, as shown in figure below. Simple graphemes have been conceived for writer verification and identification purposes. LITRP-SGBD database was made by 50 writers, who wrote 50 samples for each simple grapheme. The images are 800×800 pixels with 24-bit color and a scanning resolution of 1200 dpi.

In order to illustrate the use of some available tools to develop a CNN, following we show the implementation of two examples for fruit classification. The implementations was coded in Python and MATLAB, by using TensorFlow and the Deep Learning Toolbox, respectively.