Visual place recognition (VPR)/Loop closure detection (LCD) papers. This project mainly concerns deep learning-based methods, covering various aspects, including approaches based on Convolutional neural networks (CNNs), auto-encoders (AEs), generative adversial networks (GANs), multi-modality feature fusion, transfer learning/domain adaptation and semantic information and so forth, and challenging problems such as long-term, across seasons and heterogeneous data fusion (e.g., RGB-D and LiDAR) and etc.
- The papers will be classified acoording to their specific topics, likes thefollowing pictures.
| Index | Title | Year | Comments |
|---|---|---|---|
| 4 | CNN Features Off-the-Shelf: An Astounding Baseline for Recognition | 2014 | |
| 3 | Robust visual robot localization across seasons using network flows | 2014 | |
| 2 | Condition-invariant, top-down visual place recognition | 2014 | |
| 1 | Convolutional Neural Network-based Place Recognition | 2014 | The first CNN-based VPR method |