Mohammad Mahdi Johari

Hello! I am Mohammad Mahdi Johari, currently a Ph.D. student at EPFL, and Research Assistant in Machine Learning Group at Idiap Research Institute under the joint supervision of Prof. Fran├žois Fleuret and Prof. Daniel Gatica-Perez.

As a part of my PhD study, I did an internship at Apple focusing on fast and efficient 3D representation for human faces.

Prior to joining Idiap, I received my Master degree in Electrical Engineering from Sharif University of Technology. Thereafter, I continued my research as Research Assistant in Computer Vision Lab at Sharif University. I also worked as a Machine Learning Engineer at Cafe Bazaar.

Email  /  CV  /  Google Scholar  /  Twitter  /  LinkedIn

profile photo
Research

I am interested in Machine Learning, Computer Vision, and 3D Computer Vision.

The goal of my Ph.D. is to improve the performance of 3D reconstruction and novel view synthesis models that exploit depth information from RGBD sensors.

Publications
ESLAM: Efficient Dense SLAM System Based on Hybrid Representation of Signed Distance Fields

M. M. Johari, C. Carta, F. Fleuret
CVPR (Highlight), 2023
Paper / Code / Project Page

ESLAM is an efficient implicit neural representation method for Simultaneous Localization and Mapping (SLAM).

GeoNeRF: Generalizing NeRF With Geometry Priors

M. M. Johari, Y. Lepoittevin, F. Fleuret
CVPR , 2022
Paper / Code / Project Page

GeoNeRF is a generalizable photorealistic novel view synthesis method based on neural radiance fields.

DepthInSpace: Exploitation and Fusion of Multiple Video Frames for Structured-Light Depth Estimation

M. M. Johari, C. Carta, F. Fleuret
ICCV, 2021
Paper / Code / Project Page

DepthInSpace is a self-supervised deep-learning method for depth estimation using a structured-light camera.

Context-Aware Colorization of Gray-Scale Images Utilizing a Cycle-Consistent Generative Adversarial Network Architecture

M. M. Johari, H. Behroozi
Neurocomputing, 2020
Paper

We propose a two-stage architecture for image colorization. The first stage generates an initial colored image and selects one of the specialist networks in the second stage which improves the quality of the colored image.

Gray-Scale Image Colorization Using Cycle-Consistent Generative Adversarial Networks with Residual Structure Enhancer

M. M. Johari, H. Behroozi
ICASSP, 2020
Paper

Proposing a cycle-consistent colorization model based on Generative Adversarial Network (GAN).

Robust Airborne Target Recognition Based on Recurrence Plot Quantification of Micro-Doppler Radar Signatures

M. M. Johari, M. M. Nayebi
IRS, 2016
Paper

Proposing a method based on Recurrence Plot and Recurrence Quantification Analysis (RQA) to generate robust features against noise.


Thanks to Jon Barron for the template.