Research map / Detection and segmentation / Weakly supervised detection and segmentation
Discriminative Multiple Instance Hyperspectral Target Characterization
Alina Zare, Changzhe Jiao, Taylor C. Glenn
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| increase number of negative bags from 1 to K=15, 100, then one per background point (ablation) | EM-DD-P · MUUFL Gulfport Brown (train F1, test F3) | NAUC: 0.467 -> 0.038 -> 0.122 -> 0.420 (one bag, K=15, K=100, K=106548) | Improved |
| vary negative bag construction (one bag vs K=15/100/106548) (ablation) | MI-SMF · MUUFL Gulfport Brown (train F1, test F3) | NAUC: 0.448 -> 0.461 -> 0.454 -> 0.448 | No clear effect |
| vary negative bag construction (one bag vs K=15/100/106548) (ablation) | MI-ACE · MUUFL Gulfport Brown (train F1, test F3) | NAUC: 0.474 -> 0.496 -> 0.476 -> 0.474 | No clear effect |
| increase target proportion to 0.25 with 3 positive bags of 2 target points | MI-SMF · Simulated hyperspectral | AUC: 0.995±0.001 for MI-SMF, 0.994±0.001 for MI-ACE, 0.840±0.167 eFUMI, 0.506±0.110 EM-DD | Improved |
| raise proportion of positive bags from 0.05 to 0.25 | MI-SMF · Simulated hyperspectral | AUC: 0.838±0.303 -> 0.988±0.008 | Improved |
| raise proportion of positive bags from 0.05 to 0.25 | MI-ACE · Simulated hyperspectral | AUC: 0.716±0.368 -> 0.917±0.197 | Improved |
| raise fraction of target points in positive bags from 0.05 to 0.25 | MI-SMF · Simulated hyperspectral | AUC: 0.925±0.161 -> 0.984±0.005 | Improved |
| raise fraction of target points in positive bags from 0.05 to 0.25 | MI-ACE · Simulated hyperspectral | AUC: 0.811±0.224 -> 0.981±0.006 | Improved |
| lower mean target proportion from 0.25 to 0.05 | MI-SMF · Simulated hyperspectral | AUC: 0.989±0.002 -> 0.984±0.003 | No clear effect |
| lower mean target proportion from 0.25 to 0.05 | eFUMI · Simulated hyperspectral | AUC: 0.985±0.002 -> 0.964±0.012 | Hurt |
| estimate discriminative signature with MI-ACE instead of using true target signature | SMF/ACE detectors · Simulated 2D dataset 2 | AUC: 0.81 -> 0.86 (SMF), 0.86 -> 0.95 (ACE) | Improved |
| use MI-SMF/MI-ACE instead of eFUMI for speed | MI-SMF · Simulated hyperspectral | run time: 0.006 s vs 1.082 s average run time | Improved |
Closest papers by meaning
- Dual-stage Hyperspectral Image Classification Model with Spectral Supertoken (ECCV 2024 · similarity 0.83)
- Modeling Spectral-Spatial Attributes for Hyperspectral Anomaly Detection (TPAMI 2026 · similarity 0.83)
- HyperFree: A Channel-adaptive and Tuning-free Foundation Model for Hyperspectral Remote Sensing Imagery (CVPR 2025 · similarity 0.81)
- Learning De-Biased Representations for Remote-Sensing Imagery (NeurIPS 2024 · similarity 0.81)
- A Selective Re-learning Mechanism for Hyperspectral Fusion Imaging (CVPR 2025 · similarity 0.81)
- LESSViT: Robust Hyperspectral Representation Learning under Spectral Configuration Shift (NeurIPS 2026 · similarity 0.80)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
