Arash Amini

About

Arash Amini is a leading researcher in computer vision and robotics, specializing in 6D object pose estimation—a critical technology for autonomous robot manipulation. His most impactful work centers on the YOLOPose series, which pioneered the use of Transformer architectures for multi-object 6D pose estimation via keypoint regression. The original YOLOPose (2022) and its enhanced version, YOLOPose V2 (2023), each garnered 28 citations, demonstrating significant influence in the field. These works advanced beyond traditional CNN-based models by leveraging Transformers’ ability to capture long-range dependencies, enabling more accurate and robust pose predictions for real-world robotic applications. Amini also contributed to real-time pose estimation for humanoid robots and introduced T6D-Direct, a direct regression approach using Transformers. Beyond pose estimation, his research includes robustness analysis of recurrent neural networks under perturbed sequential inputs, quantifying stability bounds for classification tasks. His work bridges cutting-edge deep learning architectures with practical robotic systems, making him a notable figure in the intersection of computer vision and autonomous systems.

Research Focus

Key Achievements

Top Papers

1. 1
   YOLOPose: Transformer-Based Multi-object 6D Pose Estimation Using Keypoint Regression

   28 citations · 2023
2. 2
   YOLOPose V2: Understanding and improving transformer-based 6D pose estimation

   28 citations · 2023
3. 3
   Real-Time Pose Estimation from Images for Multiple Humanoid Robots

   7 citations · 2022
4. 4
   T6D-Direct: Transformers for Multi-object 6D Pose Direct Regression

   3 citations · 2021
5. 5
   YOLOPose: Transformer-based Multi-Object 6D Pose Estimation using Keypoint Regression

   3 citations · 2022
6. 6
   Robustness Analysis of Classification Using Recurrent Neural Networks with Perturbed Sequential Input

   2 citations · 2022

Key Collaborators

Contact & Links