Tung Do

Tung Do

M.Sc. Electrical and Computer Engineering - Robotics candidate | Research Assistant | Robotics Engineer

Monocular 3D Object Detection in Foggy Conditions

Published:

Advisor: Prof. Maani Ghaffari



Brief: This paper investigates improvements in monocular 3D object detection, which is crucial for autonomous vehicles. The focus is on increasing detection accuracy and robustness in various weather conditions, especially in fog. The study uses the MonoCon model, incorporating transfer learning, image augmentation, and pre-processing techniques to enhance visibility in foggy conditions. It addresses specific challenges like fluctuating Average Precision (AP) values and the inefficient detection of distant or small vehicles in fog by revising the evaluation strategy and using targeted image processing.

Role: Robotics Researcher

Result

- Led the enhancement of the MonoCon model using PyTorch, focusing on transfer learning and advanced image processing techniques to improve 3D object detection in foggy conditions. This work aimed at increasing the reliability of perception systems for autonomous robotics in challenging environments.
- Implemented and fine-tuned advanced image augmentation and pre-processing strategies to boost detection accuracy and robustness under varied weather conditions, ensuring that the system can operate effectively in real-world scenarios involving environmental uncertainties.
MonoCon Architecture by 2gunsu

Normal condition test

Foggy condition test

AP values trend over epoch trained

[GitHub][Publication]

Skills: Leadership, Machine Learning (PyTorch), Python, Linux, Bash/Shell Scripting, Git, Docker
Contributors' Acknowledgement: 2gunsu/monocon-pytorch, Minghan Zhu, Xirong Liu, Rahul Swayampakula