TopicFM: Robust and Interpretable Topic-Assisted Feature Matching

Abstract

This study addresses an image-matching problem in challenging cases, such as large scene variations or textureless scenes. To gain robustness to such situations, most previous studies have attempted to encode the global contexts of a scene via graph neural networks or transformers. However, these contexts do not explicitly represent high-level contextual information, such as structural shapes or semantic instances; therefore, the encoded features are still not sufficiently discriminative in challenging scenes. We propose a novel image-matching method that applies a topic-modeling strategy to encode high-level contexts in images. The proposed method trains latent semantic instances called topics. It explicitly models an image as a multinomial distribution of topics, and then performs probabilistic feature matching. This approach improves the robustness of matching by focusing on the same semantic areas between the images. In addition, the inferred topics provide interpretability for matching the results, making our method explainable. Extensive experiments on outdoor and indoor datasets show that our method outperforms other state-of-the-art methods, particularly in challenging cases. The code is available at https://github.com/TruongKhang/TopicFM.

Accepted at AAAI-23. This version includes both main text and supplementary materials

References in corpus (11)

• An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
• Fully Convolutional Networks for Semantic Segmentation
• R2D2: Repeatable and Reliable Detector and Descriptor
• LoFTR: Detector-Free Local Feature Matching with Transformers
• ASLFeat: Learning Local Features of Accurate Shape and Localization
• ContextDesc: Local Descriptor Augmentation with Cross-Modality Context
• Image Stylization for Robust Features
• COTR: Correspondence Transformer for Matching Across Images
• Patch2Pix: Epipolar-Guided Pixel-Level Correspondences
• Efficient Neighbourhood Consensus Networks via Submanifold Sparse Convolutions
• ClusterGNN: Cluster-based Coarse-to-Fine Graph Neural Network for Efficient Feature Matching

Cited by in corpus (2)

• OAMatcher: An Overlapping Areas-based Network for Accurate Local Feature Matching
• TopicFM+: Boosting Accuracy and Efficiency of Topic-Assisted Feature Matching