Projects...

Notable R&D Projects:

• Novel ConvNet architecture (Patched-CNN) for fast and robust Shadow Detection of natural scene images with speed (1.5sec/image) and accuracy (%91), [Code], [Project Demo]
    
• Visual Place Recognition, [Code]
   
• Vehicle License Plate Recognition, CURRENNT (CUDA-enabled machine learning library for recurrent neural networks), LSTM
   
• C++ version of whole MLFlow library to manage the ML lifecycle - experimentation, monitoring, deployment, and added the reproducibility (dockerization) feature, [Code] 
    
• Analysis of Unsupervised Feature Learning in Image Segmentation, [Code], [Paper]
    - Through extensive experiments on buildings segmentation and multiple sclerosis lesions segmentation, different parameters are compared. Discussions about the preprocessing settings, the impact of dictionary learning, encoding and classification is presented.
    
• Algorithm for Uncalibrated Visual Servoing, Numerical Methods, [Code], [Demo]
    - Algorithm for a 4-DOF robotic arm (Barrett WAM) to point at user-defined locations
    
• Hand gesture recognition method, using time-domain samples, LSTM, [Code]
    
• Algorithm for determining the most probable cause of death of a person, by reading/parsing the personal info/entities from web-page links in Wikidata, NLP, [Code]

Notable Development Projects:

• Graphical Music Player with GUI, which is capable of playing .mp3 and .ogg files, C++ and FLTK, [Code]
    
• Persian-to-English and vise versa Dictionary Software, C#, Windows Form Applications
    
• Persian Language Grammar Checker Software, [Code]
    
• Java Compiler (Javac) for Java Programming language, [Code]