Back to search

Cornell University

Analyzing Resolution Constraints in Satellite-Based Computer Vision Models

Abstract

dc:description.abstract

The increasing availability and temporal precision of satellite imagery — driven by the rapid expansion of Earth observation constellations — has significantly broadened the scope of change detection models in remote sensing. This progress has fueled applications ranging from disaster response to urban planning. However, detecting subtle and localized changes, such as demolitions of heritage sites, poses unique challenges. High-resolution imagery (0.5m) provides the necessary spatial fidelity, but remains cost-prohibitive for continuous monitoring; low-resolution (10m) imagery lacks critical detail. This thesis investigates whether medium-resolution (3m) satellite imagery can serve as a viable trade-off, enabling scalable, automated change detection through computer vision models. I adapt self-supervised change detection techniques originally developed for low-resolution imagery (e.g., Sentinel-2) to the medium-resolution setting and evaluate their ability to capture semantically meaningful, spatially subtle changes. This is paired with a supervised pipeline using a labeled dataset of medium-resolution satellite images. Through both qualitative and quantitative analyses, I assess model performance in detecting demolition-scale changes under real-world environmental variability. While medium-resolution inputs support the detection of coarse structural shifts such as new roads or land clearance, they consistently fail to capture finer details associated with building demolition — particularly in the presence of seasonal variation. These findings reveal critical resolution-bound limits in current vision pipelines and motivate future research into hybrid-resolution architectures, resolution-aware learning, and intelligent triage strategies for scalable, resource-efficient change monitoring.

Degree

thesis:*
Name thesis:degree_name
M.S., Computer Science
Level thesis:degree_level
Master of Science
Discipline thesis:degree_discipline
Computer Science
Grantor
Cornell University
Year dc:date.issued
2025

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Bhagat, Snehal
Chair dc:contributor.chair
  • Hariharan, Bharath
Committee member dc:contributor.committeemember
  • Zhang, Zhiru

Subjects

dc:subject × 6

Rights

dc:rights
Statement dc:rights
  • Attribution-NonCommercial-NoDerivatives 4.0 International
Language dc:language.iso
en

Identifiers

dc:identifier.*
Dc Identifier Bibid
17422455
Dc Identifier Other
ProQuest Submission ID: 12568
ProQuest Publication ID: 32171080

Chain of custody

source
Harvested from
Cornell University
Base URL
ecommons.cornell.edu/server/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Bhagat, Snehal. Analyzing Resolution Constraints in Satellite-Based Computer Vision Models. Master of Science thesis, Cornell University, 2025. https://hdl.handle.net/1813/120635