About Me

A machine learning engineer focused on real-time multi-stream high-performance video analysis pipelines. My primary tools are GStreamer in conjunction with computational frameworks such as Tensorflow, Pytorch, TensorRT implemented on top of specialized hardware accelerators like GPUs, TPUs, DLAs, VICs, and more.

Contact Details

Hamid Mohammadi
Iran, Tehran, Jeyhoon Ave.
(+98) 9395191963
sandstormeatwo@gmail.com
hamid.mohammadi@aut.ac.ir

Download Resume

Education

AmirKabir University of Technology

Master of Artificial Intelligence • September 2019

Thesis: Video Violence Detection using Deep Reinforcement Learning
Supervisor: Dr. Nazerfard

KhajeNasir Toosi University of Technology

Bachelor of Computer Engineering • September 2014

Thesis: Text as Environment: A Deep Reinforcement Learning Approach to Text Readability Assessment
Supervisor: Dr. Khasteh

Work

FarazPardazan

Computer Vision Engineer • Sep 2021 - Present

My focus here was to develop web and mobile friendly computer vision solutions for automated identity evaluation and optical information extraction. I am involved in projects Face Detection and Recognition, Liveness Detection, Visual Brand Identification, Credit and ID card OCR.

Sensifai

Computer Vision Engineer • July 2019 - Sep 2021

Building intelligent surveillance services, I was involved in several projects including face detection and recognition, fire detection, violence detection, object detection, motion localization, and so forth. Moreover, I participated in designing and implmenting efficient real-time hardware-accelerated multi-stream video analysis tools.

Yooz Search Engine

Junior Machine Learning and Big-Data Developer (part-time) • Augest 2016 - September 2017

My task as an artificial intelligence Engineer in this company was to help the web search team attain its goals by designing and implementing artificial intelligence solutions mainly related to Natural Language Processing and Machine learning. Some of these projects are Duplicate and near-duplicate document detection in big data applications, which mainly was about dimensionality reduction, and Persian text difficulty assessment and Persian transliteration, which were implemented in Big-Data environments.

Yooz Search Engine

Machine Learning and Big-Data Intern (part-time) • July 2016 - Augest 2016

In this summer internship, I worked on projects such as Persian sentiment analysis, Persian text similarity measurement, and offensive words filtering.

Noghteh WebServices

Android Developer (part-time) • July 2015 - July 2016

My primary goal in this company was to design and implement the ”ClassPlus” android application, an educational management, and a self-assessment assistant.

Skills

Projects

GStreamer C++

Useful GStreamer samples in C++ • GStreamer, OpenCV-C++, C++ • private, available upon request

These samples are intended for GStreamer C++ learners and bootstrapping C++ GStreamer-based products.

DeepStream C++

Useful DeepStream samples in C++ • DeepStream, GStreamer, OpenCV-C++, OpenCV-CUDA, C++ • private, available upon request

These samples are intended for Nvidia DeepStream learners and GStreamer C++ and bootstrapping C++ DeepStream-based products.

C++/Python interface

Various Pybind11 examples • Pybind11, C++, Python •

A collection of Pybind11 library samples. Pybind11 is a library aimed at implementing Python modules in C++ language. These examples are useful for learning and bootstrapping projects.

SharkTop

Curses-based user interface for GstShark • Python, Curses •

GstShark is a comprehensive debug tool for GStreamer. Sharktop displays debug information by launching a pipeline using GstShark or attaching to a currently running GStreamer + GstShark pipeline.

Tile Player

GStreamer-based multi-stream tiled player • GStreamer, Python, Bash •

Simply display multiple video files or streams using a configurable and easy to use linux command tool.

Simple Centroid Tracker

A light-weight centroid tracker • OpenCV, Python, NumPy •

High-performance low-accuracy tracking for use cases with limited processing power and tolerance for tracking errors.

GStreamer python samples

Useful GStreamer samples in Python • GStreamer, Python • private, under development

These samples are intended for GStreamer Python learners and bootstrapping GStreamer Python based projects.

Sci-kit learn samples

Useful Sci-kit learn samples in Python • Sci-kit learn, Python • private, under development

These samples are intended for Sci-kit learners and bootstrapping Sci-kit learn based projects.

Keras samples

Useful Keras (Tensorflow 2.x) samples in Python • Keras, Tensorflow 2.x, Python • private, under development

These samples are intended for Keras and tensorflow learners and bootstrapping Keras (Tensorflow 2.x) projects.

Research

Hamid Mohammadi, Ehsan Nazerfard, "SSHA: Video Violence Recognition and Localization using a Semi-Supervised Hard-Attention Model

Under review • PDF

Current human-based surveillance systems are prone to inadequate availability and reliability. Artificial intelligence-based solutions are compelling, considering their reliability and precision in the face of an increasing adaption of surveillance systems. The proposed model uses an I3D backbone pretrained on the Kinetics dataset and has achieved state-of-the-art accuracy of 90.4% and 98.7% on RWF and Hockey datasets, respectively. The semi-supervised hard attention mechanism has enabled the proposed method to fully capture the available information in a high-resolution video by processing the necessary video regions in great detail.

Hamid Mohammadi, Seyed Hossein Khasteh, ”Text as Environment: A Deep Reinforcement Learning Text Readability Assessment Model”

Preprint • PDF

Deep reinforcement learning models are demonstrated to be helpful in further improvement of stateof-the-art text readability assessment models. The main contributions of the proposed approach are the automation of feature extraction, loosening the tight language dependency of text readability assessment task, and efficient use of text by finding the minimum portion of a text required to assess its readability. The experiments on Weebit, Cambridge Exams, and Persian readability datasets display the model’s state-of-theart precision, efficiency, and the capability to be applied to other languages.

Hamid Mohammadi, Seyed Hossein Khasteh, Amin Nikoukaran, ”A Machine Learning Approach to Persian Text Readability Assessment Using a Crowdsourced Dataset”

28th Iranian Conference on Electrical Engineering (ICEE) • PDF

In the present research, the first Persian dataset for text readability assessment was gathered and the first model for Persian text readability assessment using machine learning was introduced. The experiments showed that this model was accurate and could assess the readability of Persian texts with a high degree of confidence.

Hamid Mohammadi, Seyed Hossein Khasteh, ”A Fast Text Similarity Measure for Large Document Collections Using Multi-reference Cosine and Genetic Algorithm”

Published at Turkish Journal of Electrical Engineering Computer Sciences • PDF

In this paper, a new signature-based approach to text similarity detection is introduced which is fast, scalable, reliable and needs less storage space. The proposed method is examined on popular text document data-sets such as CiteseerX, Enron, Gold Set of Near-duplicate News Articles and etc. The results are promising and comparable with the best cutting-edge algorithms, considering the accuracy and performance.

Hamid Mohammadi, Amin Nikoukaran, ”Multi-reference Cosine: A New Approach to Text Similarity Measurement in Large Collections”

Preprint • PDF

In this paper, a new signature-based approach to text similarity detection is introduced which is fast, scalable, reliable and needs less storage space. The proposed method is examined on popular text document data-sets such as CiteseerX, Enron, Gold Set of Near-duplicate News Articles and etc. The results are promising and comparable with the best cutting-edge algorithms, considering the accuracy and performance.