PhD student at Intelligent Human Perception Lab, USC Institute for Creative Technologies
University of Southern California, Thomas Lord Department of Computer Science, USC Viterbi School of Engineering,
Intelligent Human Perception Lab
2023–current

Bachelor of Computer Science, Moscow State University named after M. V. Lomonosov
Applied Mathematics and Computer Science, Department of Intelligent Information Technologies,
Graphics and Media Lab
2019–2023

SEMPI: A Database for Understanding Social Engagement in Video-Mediated Multiparty Interaction
Maksim Siniukov*, Yufeng Yin*, Eli Fast, Yingshan Qi, Aarav Monga, Audrey Kim, Mohammad Soleymani
Accepted to ACM International Conference on Multimodal Interaction (ICMI) 2024

Dyadic Interaction Modeling for Social Behavior Generation
Minh Tran*, Di Chang*, Maksim Siniukov, Mohammad Soleymani
Accepted to European Conference on Computer Vision (ECCV) 2024
[PDF]

Unveiling the Limitations of Novel Image Quality Metrics
Maksim Siniukov, Dmitriy Kulikov, Dmitriy Vatolin
2023 IEEE 25th International Workshop on Multimedia Signal Processing (MMSP) 2023
[PDF]

Hacking VMAF and VMAF NEG: vulnerability to different preprocessing methods
Maksim Siniukov, Anastasia Antsiferova, Dmitriy Kulikov, Dmitriy Vatolin
AICCC’21: 2021 4th Artificial Intelligence and Cloud Computing Conference
[PDF]

Applicability limitations of differentiable full-reference image-quality metrics
Maksim Siniukov, Dmitriy Kulikov, Dmitriy Vatolin
Data Compression Conference (DCC) 2023
[arXiv]

NETFLIX VMAF no enchantment gain vulnerability to sharpness and contrast transformations
Maksim Siniukov, Anastasia Antsiferova
International Scientific Conference Lomonosov-2021

Limitations of applicability of differentiable reference indicators of image quality
Maksim Siniukov, Dmitry Kulikov, Dmitry Vatolin, Vladimir Galaktionov
Journal "IPM named after M.V. Keldysh", 2022

Hacking VMAF and VMAF NEG: vulnerability to different preprocessing methods
Asia Digital Image Processing Conference 2021

NETFLIX VMAF no enchantment gain vulnerability to sharpness and contrast transformations
International youth scientific conference Lomonosov-2021

High-frequency high-voltage short time electric discharge in air
The 39-th Beijing Young Science Creation Competition

The First Prize of Excellent Youth Science & Technology Innovation Project, The 39-th Beijing Young Science Creation Competition, first place

The 2-nd place at the competition of scientific and technical schoolchildren works ”Scientists of the Future”, 11-th grade, MSU, 2018

The 3-rd place in the All-Russian competition of schoolchildren scientific works ”Junior”, 9-th grade, MEPhI, 2017

Olympiad ”Kurchatov”, prize-winner

Olympiad ”Phystech”, winner

Engineering Olympiad for schoolchildren, winner

Olympiad of St. Petersburg State University, prize-winner

Stanford University Machine Learning Course
Stanford Online
[Certificate]

Specialization "Deep Learning Specialization"
DeepLearning.AI
[Certificate]

Course Convolutional Neural Networks
Coursera
[Certificate]

Course Neural Networks and Deep Learning
DeepLearning.AI.
[Certificate]

Course Structuring Machine Learning Projects
Coursera
[Certificate]

Course Sequence Models
Coursera
[Certificate]

Course Improving Deep Neural Networks: Hyperparameter tuning, Regularization and Optimization
DeepLearning.AI
[Certificate]

Educational course on the basics of sports programming at the MISiS Research Technological University, 2018
National University of Science and Technology (MISiS)

English Language Certificate: LTC. General English. Certificate of Attendance, 2018
LTC

Course "Media data processing and compression methods"
CMC MSU

Course "Intelligent methods of video processing"
CMC MSU

Programming:
Python(PyTorch, Tensorflow 2, OpenCV, CatBoost, pandas, CUDA OpenCV, SciPy, DEAP, SymPy, Keras, PyTorch Lightning, scikit-learn, numba), C++, C, MATLAB(Octave), NASM, Java
Technical skills:
Docker, git, ssh, sanitizers, Valgrind, profilers, Make, ffmpeg, VQMT, LaTeX
Languages:
Russian(Native), English(Advanced, IELTS: 7.0/9.0, passed at October 2022)

Supervisors: Dmitry Vatolin, Dmitriy Kulikov, Anastasia Antsiferova
Description: Video-quality measurement plays a critical role in the development of video-processing applications. In this regard, more and more metrics are under development, but little research has considered their limitations. In this paper, we show how video preprocessing can artificially increase the popular quality metric VMAF and its tuning-resistant version, VMAF NEG. We propose a pipeline that tunes processing-algorithm parameters to increase VMAF by up to 218.8%. A subjective comparison revealed that for most preprocessing methods, a video’s visual quality drops or stays unchanged. We also show that some preprocessing methods can increase VMAF NEG scores by up to 23.6%. We show how image preprocessing before compression can artificially increase the quality scores provided by the popular metrics DISTS, LPIPS, HaarPSI, and VIF as well as how these scores are inconsistent with subjective- quality scores. We propose a series of neural-network preprocessing models that increase DISTS by up to 34.5%, LPIPS by up to 36.8%, VIF by up to 98.0%, and HaarPSI by up to 22.6% in the case of JPEG-compressed images. A subjective comparison of preprocessed images showed that for most of the metrics we examined, visual quality drops or stays unchanged, limiting the applicability of these metrics