# Do Convnets Learn Correspondence?

arXiv: https://arxiv.org/abs/1411.1091 · HTML: https://mlsanity.com/paper/1411.1091/

## Key facts

- Cited by 28 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 4 Nov 2014.

## Built on by (most cited first, 28 of 28)

- [Rethinking Space-Time Networks with Improved Memory Coverage for Efficient Video Object Segmentation](https://mlsanity.com/paper/2106.05210/) (2021)
- [SSTVOS: Sparse Spatiotemporal Transformers for Video Object Segmentation](https://mlsanity.com/paper/2101.08833/) (2021)
- [Do Different Tracking Tasks Require Different Appearance Models?](https://mlsanity.com/paper/2107.02156/) (2021)
- [Neural Texture Extraction and Distribution for Controllable Person Image Synthesis](https://mlsanity.com/paper/2204.06160/) (2022)
- [Deep Lucas-Kanade Homography for Multimodal Image Alignment](https://mlsanity.com/paper/2104.11693/) (2021)
- [AlignMixup: Improving Representations By Interpolating Aligned Features](https://mlsanity.com/paper/2103.15375/) (2021)
- [Zero-Shot Everything Sketch-Based Image Retrieval, and in Explainable Style](https://mlsanity.com/paper/2303.14348/) (2023)
- [MIDMs: Matching Interleaved Diffusion Models for Exemplar-based Image Translation](https://mlsanity.com/paper/2209.11047/) (2022)
- [Bi-level Feature Alignment for Versatile Image Translation and Manipulation](https://mlsanity.com/paper/2107.03021/) (2021)
- [Distillation of Diffusion Features for Semantic Correspondence](https://mlsanity.com/paper/2412.03512/) (2024)
- [Semantic-Aware Fine-Grained Correspondence](https://mlsanity.com/paper/2207.10456/) (2022)
- [HumanGPS: Geodesic PreServing Feature for Dense Human Correspondences](https://mlsanity.com/paper/2103.15573/) (2021)
- [NeuroCartography: Scalable Automatic Visual Summarization of Concepts in Deep Neural Networks](https://mlsanity.com/paper/2108.12931/) (2021)
- [Demystifying Unsupervised Semantic Correspondence Estimation](https://mlsanity.com/paper/2207.05054/) (2022)
- [Self-Supervised Learning of Image Scale and Orientation](https://mlsanity.com/paper/2206.07259/) (2022)
- [SAME++: A Self-supervised Anatomical eMbeddings Enhanced medical image registration framework using stable sampling and regularized transformation](https://mlsanity.com/paper/2311.14986/) (2023)
- [Semantic keypoint-based pose estimation from single RGB frames](https://mlsanity.com/paper/2204.05864/) (2022)
- [Transfer of Representations to Video Label Propagation: Implementation Factors Matter](https://mlsanity.com/paper/2203.05553/) (2022)
- [Visual Analytics of Neuron Vulnerability to Adversarial Attacks on Convolutional Neural Networks](https://mlsanity.com/paper/2303.02814/) (2023)
- [Semantic Correspondence: Unified Benchmarking and a Strong Baseline](https://mlsanity.com/paper/2505.18060/) (2025)
- [DeepVerge: Classification of Roadside Verge Biodiversity and Conservation Potential](https://mlsanity.com/paper/2206.04271/) (2022)
- [Image Keypoint Matching using Graph Neural Networks](https://mlsanity.com/paper/2205.14275/) (2022)
- [Geo-NI: Geometry-aware Neural Interpolation for Light Field Rendering](https://mlsanity.com/paper/2206.09736/) (2022)
- [Real-time Detection of 2D Tool Landmarks with Synthetic Training Data](https://mlsanity.com/paper/2210.11991/) (2022)
- [Introducing the structural bases of typicality effects in deep learning](https://mlsanity.com/paper/2107.03279/) (2021)
- [Learning Space-Time Semantic Correspondences](https://mlsanity.com/paper/2306.10208/) (2023)
- [Differentiable Proximal Graph Matching](https://mlsanity.com/paper/2405.16479/) (2024)
- [Gromov Wasserstein Optimal Transport for Semantic Correspondences](https://mlsanity.com/paper/2602.03105/) (2026)

## Abstract (arXiv)

Convolutional neural nets (convnets) trained from massive labeled datasets have substantially improved the state-of-the-art in image classification and object detection. However, visual understanding requires establishing correspondence on a finer level than object category. Given their large pooling regions and training from whole-image labels, it is not clear that convnets derive their success from an accurate correspondence model which could be used for precise localization. In this paper, we study the effectiveness of convnet activation features for tasks requiring correspondence. We present evidence that convnet features localize at a much finer scale than their receptive field sizes, that they can be used to perform intraclass alignment as well as conventional hand-engineered features, and that they outperform conventional features in keypoint prediction on objects from PASCAL VOC 2011.