Aleksandar (Alex) Vakanski


University of Idaho – Physical Rehabilitation Movement Dataset (link)

UI-PRMD is a data set of movements related to common exercises performed by patients in physical therapy and rehabilitation programs. The data set consists of 10 rehabilitation exercises. A sample of 10 healthy individuals repeated each exercise 10 times in front of two sensory systems for motion capturing: a Vicon optical tracker, and a Kinect camera. The data is presented as positions and angles of the body joints in the skeletal models provided by the Vicon and Kinect mocap systems.


Codes for Attention-Enriched Deep Learning Model for Breast Tumor Segmentation (link)

The codes provide the implementation of our work on breast tumor segmentation using a salient attention deep learning model. Our proposed approach integrates prior knowledge of visual saliency into a deep learning model for tumor segmentation in ultrasound images. Visual saliency refers to image maps containing regions that are more likely to attract radiologists’ visual attention. The model introduces attention blocks into a U-Net architecture and learns feature representations that prioritize spatial regions with high saliency levels. The validation results indicate increased accuracy for tumor segmentation relative to models without salient attention layers.


Codes for A Deep Learning Framework for Assessing Physical Rehabilitation Exercises (link)

The codes are based on the research project A Deep Learning Framework for Assessing Physical Rehabilitation Exercises. The framework for automated quality assessment of physical rehabilitation exercises encompasses metrics for quantifying movement performance, scoring functions for mapping the performance metrics into numerical scores of movement quality, techniques for dimensionality reduction, and deep neural network models for regressing quality scores of input movements via supervised learning. The proposed framework employs an autoencoder network for dimensionality reduction, a performance metric based on the log-likelihood of a Gaussian mixture model, and a deep convolutional neural network for movement assessment. The spatio-temporal neural network arranges data into temporal pyramids, and exploits the spatial characteristics of human movements by using dedicated sub-networks for processing the joint displacements of individual body parts.

A reproducible version of the codes is also published on Code Ocean and can be accessed via the following link: The embedded interactive capsule widget below allows access to the Files and the Results via the Reproducibility tab.



A Dataset of Multispectral Potato Plants Images (link)

The dataset contains aerial agricultural images of a potato field with manual labels of healthy and stressed plant regions. The images were collected with a Parrot Sequoia multispectral camera carried by a 3DR Solo drone flying at an altitude of 3 meters. The dataset consists of RGB images with a resolution of 750×750 pixels, and spectral monochrome red, green, red-edge, and near-infrared images with a resolution of 416×416 pixels, and XML files with annotated bounding boxes of healthy and stressed potato crop.


Codes for Rehabilitation Assessment through Dimensionality Reduction and Statistical Modeling (link)

Codes for the paper Williams et al. (2019) use a method for evaluation of the consistency of human movements within the context of physical therapy and rehabilitation. Captured movement data in the form of joint angular displacements in a skeletal human model are used. The proposed approach employs an autoencoder neural network to project the high-dimensional movement trajectories into a low-dimensional manifold. Afterwards, a Gaussian mixture model is used to derive a parametric probabilistic model of the density of low-dimensional embeddings. The resulting probabilistic model is employed for evaluation of the consistency of unseen movement sequences based on the likelihood of the data being drawn from the model.

A reproducible version of the codes is published on Code Ocean and can be accessed via the following link:


MATLAB codes for Visual Servoing (download link 0.1 MB)

The toolbox contains MATLAB codes for image-based (IBVS) and position-based (PBVS) visual servoing. Eye-in-hand configuration of the camera has been considered. The codes are based on the Visual Servoing Toolbox for MATLAB/Simulink [E. Cervera, "Visual servoing toolbox for MATLAB/Simulink," 2003, available online:]. The toolbox transferred the Simulink models into MATLAB codes. I used some of the functions from the Visual Servoing Toolbox for MATLAB/Simulink (e.g., camera, ht, polyhedra), and I modified some of the existing functions. The codes provided here work independently, i.e., you don't need to install the Visual Servoing Toolbox for MATLAB/Simulink in order to run them. Please let me know if you can't run the codes, if you find errors, or if you have any questions regarding the codes.