HyNet: Learning Local Descriptor with Hybrid Similarity Measure and Triplet Loss

Authors: Yurun Tian, Axel Barroso Laguna, Tony Ng, Vassileios Balntas, Krystian Mikolajczyk

NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Hy Net surpasses previous methods by a significant margin on standard benchmarks that include patch matching, verification, and retrieval, as well as outperforming full end-to-end methods on 3D reconstruction tasks.
Researcher Affiliation Collaboration 1 Imperial College London 2 Facebook Reality Labs
Pseudocode No The paper describes the network architecture and mathematical formulations, but does not include any explicitly labeled 'Pseudocode' or 'Algorithm' blocks, nor does it present structured steps in a code-like format.
Open Source Code Yes Codes and models are available at https://github.com/yuruntian/Hy Net.
Open Datasets Yes UBC dataset [7] consists of three subset-scenes, namely Liberty, Notredame and Yosemite.
Dataset Splits Yes Following the evaluation protocol [7], models are trained on one subset and tested on the other two.
Hardware Specification No The paper describes training parameters (e.g., epochs, batch size, optimizer) but does not provide specific details regarding the hardware used for running the experiments (e.g., GPU model, CPU type, memory).
Software Dependencies No Our novel architecture and training is implemented in Py Torch [32].
Experiment Setup Yes The network is trained for 200 epochs with a batch size of 1024 and Adam optimizer [20]. Training starts from scratch, and the threshold τ in TLU for each layer is initialised with 1. We set α = 2 and γ = 0.1.