mlsanityData to 13 Oct 2026

Disentangled Latent Transformer for Interpretable Monocular Height Estimation

arXiv 2201.06357 ↗First posted 17 Jan 2022cs.CVcited by 3

Not on the map. Why?

Abstract (arXiv)

Monocular height estimation (MHE) from remote sensing imagery has high potential in generating 3D city models efficiently for a quick response to natural disasters. Most existing works pursue higher performance. However, there is little research exploring the interpretability of MHE networks. In this paper, we target at exploring how deep neural networks predict height from a single monocular image. Towards a comprehensive understanding of MHE networks, we propose to interpret them from multiple levels: 1) Neurons: unit-level dissection. Exploring the semantic and height selectivity of the learned internal deep representations; 2) Instances: object-level interpretation. Studying the effects of different semantic classes, scales, and spatial contexts on height estimation; 3) Attribution: pixel-level analysis. Understanding which input pixels are important for the height estimation. Based on the multi-level interpretation, a disentangled latent Transformer network is proposed towards a more compact, reliable, and explainable deep model for monocular height estimation. Furthermore, a novel unsupervised semantic segmentation task based on height estimation is first introduced in this work. Additionally, we also construct a new dataset for joint semantic segmentation and height estimation. Our work provides novel insights for both understanding and designing MHE models.

Key facts. Cited by 3 papers in the map (data through 13 Oct 2026). Builds on 29 of its 42 references that are in the map. First posted on arXiv 17 Jan 2022.

Builds on

29 of 42 references.

References · 29

  1. 2014Adam: A Method for Stochastic Optimizationcited by 37,093
  2. 2021Swin Transformer: Hierarchical Vision Transformer using Shifted Windowscited by 12,111
  3. 2016Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localizationcited by 6,713
  4. 2013Auto-Encoding Variational Bayescited by 4,024
  5. 2017Axiomatic Attribution for Deep Networkscited by 3,368
  6. 2013Visualizing and Understanding Convolutional Networkscited by 2,282
  7. 2017Decoupled Weight Decay Regularizationcited by 1,788
  8. 2014Depth Map Prediction from a Single Image using a Multi-Scale Deep Networkcited by 1,269
  9. 2018This Looks Like That: Deep Learning for Interpretable Image Recognitioncited by 735
  10. 2018Deep Ordinal Regression Network for Monocular Depth Estimationcited by 722
  11. 2018Invariant Information Clustering for Unsupervised Image Classification and Segmentationcited by 455
  12. 2019Enforcing geometric constraints of virtual normal for depth predictioncited by 234
17 more
  1. 2021Unsupervised Semantic Segmentation by Contrasting Object Mask Proposalscited by 190
  2. 2021PiCIE: Unsupervised Semantic Segmentation using Invariance and Equivariance in Clusteringcited by 163
  3. 2017Interpretable Convolutional Neural Networkscited by 161
  4. 2020Interpreting Super-Resolution Networks with Local Attribution Mapscited by 136
  5. 2017Deep learning in remote sensing: a reviewcited by 128
  6. 2017Semi-Supervised Deep Learning for Monocular Depth Map Predictioncited by 109
  7. 2018On the importance of single directions for generalizationcited by 99
  8. 2019How do neural networks see depth in single images?cited by 82
  9. 2020The Edge of Depth: Explicit Constraints between Segmentation and Depthcited by 72
  10. 2018Semantic Stereo for Incidental Satellite Imagescited by 53
  11. 2018IM2HEIGHT: Height Estimation from Single Monocular Imagery via Fully Residual Convolutional-Deconvolutional Networkcited by 31
  12. 2020Training Interpretable Convolutional Neural Networks by Differentiating Class-specific Filterscited by 26
  13. 2019Visualization of Convolutional Neural Networks for Monocular Depth Estimationcited by 24
  14. 2020Learning Geocentric Object Pose in Oblique Monocular Imagescited by 16
  15. 2021Towards Interpretable Deep Networks for Monocular Depth Estimationcited by 12
  16. 2021Single View Geocentric Pose in the Wildcited by 9
  17. 2021THE Benchmark: Transferable Representation Learning for Monocular Height Estimationcited by 8

Built on by

Cited by year (2026 to 13 Oct)

Citing papers per year: 2023: 2; 2024: 0; 2025: 0; 2026: 1.

3 papers.

  1. 2023HTC-DC Net: Monocular Height Estimation from Single Remote Sensing Imagescited by 8
  2. 2023Multilingual Augmentation for Robust Visual Question Answering in Remote Sensing Imagescited by 7
  3. 2026Reconstructing Building Height from Spaceborne TomoSAR Point Clouds Using a Dual-Topology Networkcited by 2