Saved in:
Bibliographic Details
Main Authors: Govindarajan, Hariprasath, Wozniak, Maciej K., Klingner, Marvin, Maurice, Camille, Kiran, B Ravi, Yogamani, Senthil
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2503.09878
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912722655903744
author Govindarajan, Hariprasath
Wozniak, Maciej K.
Klingner, Marvin
Maurice, Camille
Kiran, B Ravi
Yogamani, Senthil
author_facet Govindarajan, Hariprasath
Wozniak, Maciej K.
Klingner, Marvin
Maurice, Camille
Kiran, B Ravi
Yogamani, Senthil
contents Vision foundation models (VFMs) such as DINO have led to a paradigm shift in 2D camera-based perception towards extracting generalized features to support many downstream tasks. Recent works introduce self-supervised cross-modal knowledge distillation (KD) as a way to transfer these powerful generalization capabilities into 3D LiDAR-based models. However, they either rely on highly complex distillation losses, pseudo-semantic maps, or limit KD to features useful for semantic segmentation only. In this work, we propose CleverDistiller, a self-supervised, cross-modal 2D-to-3D KD framework introducing a set of simple yet effective design choices: Unlike contrastive approaches relying on complex loss design choices, our method employs a direct feature similarity loss in combination with a multi layer perceptron (MLP) projection head to allow the 3D network to learn complex semantic dependencies throughout the projection. Crucially, our approach does not depend on pseudo-semantic maps, allowing for direct knowledge transfer from a VFM without explicit semantic supervision. Additionally, we introduce the auxiliary self-supervised spatial task of occupancy prediction to enhance the semantic knowledge, obtained from a VFM through KD, with 3D spatial reasoning capabilities. Experiments on standard autonomous driving benchmarks for 2D-to-3D KD demonstrate that CleverDistiller achieves state-of-the-art performance in both semantic segmentation and 3D object detection (3DOD) by up to 10% mIoU, especially when fine tuning on really low data amounts, showing the effectiveness of our simple yet powerful KD strategy
format Preprint
id arxiv_https___arxiv_org_abs_2503_09878
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle CleverDistiller: Simple and Spatially Consistent Cross-modal Distillation
Govindarajan, Hariprasath
Wozniak, Maciej K.
Klingner, Marvin
Maurice, Camille
Kiran, B Ravi
Yogamani, Senthil
Computer Vision and Pattern Recognition
Artificial Intelligence
Robotics
Vision foundation models (VFMs) such as DINO have led to a paradigm shift in 2D camera-based perception towards extracting generalized features to support many downstream tasks. Recent works introduce self-supervised cross-modal knowledge distillation (KD) as a way to transfer these powerful generalization capabilities into 3D LiDAR-based models. However, they either rely on highly complex distillation losses, pseudo-semantic maps, or limit KD to features useful for semantic segmentation only. In this work, we propose CleverDistiller, a self-supervised, cross-modal 2D-to-3D KD framework introducing a set of simple yet effective design choices: Unlike contrastive approaches relying on complex loss design choices, our method employs a direct feature similarity loss in combination with a multi layer perceptron (MLP) projection head to allow the 3D network to learn complex semantic dependencies throughout the projection. Crucially, our approach does not depend on pseudo-semantic maps, allowing for direct knowledge transfer from a VFM without explicit semantic supervision. Additionally, we introduce the auxiliary self-supervised spatial task of occupancy prediction to enhance the semantic knowledge, obtained from a VFM through KD, with 3D spatial reasoning capabilities. Experiments on standard autonomous driving benchmarks for 2D-to-3D KD demonstrate that CleverDistiller achieves state-of-the-art performance in both semantic segmentation and 3D object detection (3DOD) by up to 10% mIoU, especially when fine tuning on really low data amounts, showing the effectiveness of our simple yet powerful KD strategy
title CleverDistiller: Simple and Spatially Consistent Cross-modal Distillation
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Robotics
url https://arxiv.org/abs/2503.09878