Hi, I’m Mohammad Amin Shabani

PhD student - Simon Fraser University

I am a PhD student in Computer Science at Simon Fraser University, advised by Prof. Yasutaka Furukawa. I earned my MS in Electrical and Computer Engineering from Seoul National University, where I was supervised by Prof. Kyoung Mu Lee, and my BS in Computer Science from Sharif University of Technology.
My research focuses on Computer Vision, Deep Learning, and Graph Theory.

I am currently exploring full-time career opportunities in the industry. If there are any opportunities available, please feel free to contact me.

amin shabani

News


  • One paper got accepted at CVPR 2024. The work was done during my internship at Adobe. New
  • One paper got accepted at NeurIPS 2023 (Spotlight) New
  • Joined Adobe as a Research Intern, June 2023
  • One paper got accepted at CVPR 2023
  • One paper got accepted to ICLR 2023. The work was done mostly during my internship at BorealisAI.

Publications


Visual Layout Composer: Image-Vector Dual Diffusion Model for Design Layout Generation

This paper proposes an image-vector dual diffusion model for generative layout design. Distinct from prior efforts that mostly ignores visual information of elements and the whole canvas, our approach integrates the power of a pre-trained large image diffusion model to guide layout composition in a vector diffusion model by providing enhanced salient region understanding and high-level interelement relationship reasoning.

Mohammad Amin Shabani , Zhaowen Wang, Difan Liu, Nanxuan Zhao, Jimei Yang, Yasutaka Furukawa
CVPR 2024

Visual Layout Composer

PuzzleFusion: Unleashing the Power of Diffusion Models for Spatial Puzzle Solving

This paper presents an end-to-end neural architecture based on Diffusion Models for spatial puzzle solving, particularly jigsaw puzzle and room arrangement tasks.

Sepidehsadat Hosseini, Mohammad Amin Shabani , Saghar Irandoust, and Yasutaka Furukawa
NeurIPS 2023 (Spotlight)

JigsawPlan

HouseDiffusion: Vector Floorplan Generation via a Diffusion Model with Discrete and Continuous Denoising

The paper presents a new method for generating vector floorplans using a diffusion model that improves upon existing methods. The approach uses a graph-conditioned floorplan generation approach with a Transformer architecture and outperforms state-of-the-art methods on the RPLAN dataset.

Mohammad Amin Shabani, Sepidehsadat Hosseini, and Yasutaka Furukawa
CVPR 2023

HouseDiffusion

ScaleFormer: Iterative Multi-Scale Refining Transformers for Time Series Forecasting

In this paper, we propose a general multi-scale framework that can be applied to the state-of-the-art transformer-based time series forecasting models (FEDformer, Autoformer, etc.).

Mohammad Amin Shabani, Amir Abdi, Lili Meng, and Tristan Sylvain
ICLR 2023

Scaleformer

Vectorizing Building Blueprints

This paper proposes a novel vectorization algorithm for highdefinition floorplans with construction-level intricate architectural details, namely a blueprint.

Weilian Song, Mahsa Maleki Abyaneh, Mohammad Amin Shabani, and Yasutaka Furukawa
ACCV 2022

Vectorizing Building Blueprints

Extreme Structure from Motion for Indoor Panoramas without Visual Overlaps

This paper proposes an extreme Structure from Motion (SfM) algorithm for residential indoor panoramas that have little to no visual overlaps. Only a single panorama is present in a room for many cases, making the task infeasible for existing SfM algorithms.

Mohammad Amin Shabani, Weilian Song, Makoto Odamaki, Hirochika Fujiki, and Yasutaka Furukawa
ICCV 2021

extreme_sfm

Local Visual Microphones: Improved Sound Extraction from Silent Video

In this paper we study local vibration patterns at different image locations. We show that different locations in the image vibrate differently. We carefully aggregate local vibrations and produce a sound quality that improves state-of-the-art.

MA Shabani, L Samadfam, MA Sadeghi
BMVC 2017

Local Visual Microphones

Layer-wise Progressive Knowledge Distillation

We proposed a new method to create soft targets in different levels of complexity by obtaining the probabilities from the intermediate layers of the teacher network. Our method is specially designed for the cases that there is a large gap between the teacher and the student which makes it harder for the student to mimic the teacher.

MA Shabani
2019

Layer-wise Progressive Knowledge Distillation

On decomposing complete tripartite graphs into 5-cycles

The problem of finding necessary and sufficient conditions to decompose a complete tripartite graph into 5-cycles was first considered by ES Mahmoodian and Maryam Mirzakhani (1995). They stated some necessary conditions and conjectured that those conditions are also sufficient. We show the conjecture is true under some conditions.

M Abdolmaleki, SG Ilchi, ES Mahmoodian, MA Shabani
arXiv preprint

decomposing complete tripartite graphs into 5-cycles

On Uniquely k-List Colorable Planar Graphs, Graphs on Surfaces, and Regular Graphs

A graph G is said to have property M(k) if it is not uniquely k-list colorable. We found bounds on M(k) for graphs embedded on surfaces, and obtain new results on planar graphs. We begin a general study of bounds on M(k) for regular graphs, as well as for graphs with varying list sizes.

M Abdolmaleki, J. P. Hutchinson, S. Gh. Ilchi, E. S. Mahmoodian, N Matsumoto, M. A. Shabani
Graphs and Combinatorics, 2018

UkLC