Label-Efficient Hyperspectral Image Classification via Spectral FiLM Modulation of Low-Level Pretrained Diffusion Features

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Hu, Yuzhen, Banerjee, Biplab, Prasad, Saurabh
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911299761340416
author Hu, Yuzhen
Banerjee, Biplab
Prasad, Saurabh
author_facet Hu, Yuzhen
Banerjee, Biplab
Prasad, Saurabh
contents Hyperspectral imaging (HSI) enables detailed land cover classification, yet low spatial resolution and sparse annotations pose significant challenges. We present a label-efficient framework that leverages spatial features from a frozen diffusion model pretrained on natural images. Our approach extracts low-level representations from high-resolution decoder layers at early denoising timesteps, which transfer effectively to the low-texture structure of HSI. To integrate spectral and spatial information, we introduce a lightweight FiLM-based fusion module that adaptively modulates frozen spatial features using spectral cues, enabling robust multimodal learning under sparse supervision. Experiments on two recent hyperspectral datasets demonstrate that our method outperforms state-of-the-art approaches using only the provided sparse training labels. Ablation studies further highlight the benefits of diffusion-derived features and spectral-aware fusion. Overall, our results indicate that pretrained diffusion models can support domain-agnostic, label-efficient representation learning for remote sensing and broader scientific imaging tasks.
format Preprint
id arxiv_https___arxiv_org_abs_2512_03430
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Label-Efficient Hyperspectral Image Classification via Spectral FiLM Modulation of Low-Level Pretrained Diffusion Features
Hu, Yuzhen
Banerjee, Biplab
Prasad, Saurabh
Computer Vision and Pattern Recognition
Hyperspectral imaging (HSI) enables detailed land cover classification, yet low spatial resolution and sparse annotations pose significant challenges. We present a label-efficient framework that leverages spatial features from a frozen diffusion model pretrained on natural images. Our approach extracts low-level representations from high-resolution decoder layers at early denoising timesteps, which transfer effectively to the low-texture structure of HSI. To integrate spectral and spatial information, we introduce a lightweight FiLM-based fusion module that adaptively modulates frozen spatial features using spectral cues, enabling robust multimodal learning under sparse supervision. Experiments on two recent hyperspectral datasets demonstrate that our method outperforms state-of-the-art approaches using only the provided sparse training labels. Ablation studies further highlight the benefits of diffusion-derived features and spectral-aware fusion. Overall, our results indicate that pretrained diffusion models can support domain-agnostic, label-efficient representation learning for remote sensing and broader scientific imaging tasks.
title Label-Efficient Hyperspectral Image Classification via Spectral FiLM Modulation of Low-Level Pretrained Diffusion Features
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2512.03430