Efficient Object Instance Search Using Fuzzy Objects Matching

Authors: Tan Yu, Yuwei Wu, Sreyasee Bhattacharjee, Junsong Yuan

AAAI 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental The experimental results demonstrate that the proposed FOM framework significantly outperforms the state-of-the-art methods in precision with less memory and computational cost on three public datasets.
Researcher Affiliation Academia Tan Yu,1 Yuwei Wu,1,2 Sreyasee Bhattacharjee,1 Junsong Yuan1 1Rapid Rich Object Search Lab, Interdisciplinary Graduate School, Nanyang Technological Univeristy, Singapore, 637553 2Beijing Laboratory of Intelligent Information Technology, School of Computer Science, Beijing Institute of Technology, Beijing, 100081 {tyu008, dbhattacharjee, jsyuan}@ntu.edu.sg, wuyuwei@bit.edu.cn
Pseudocode No The paper does not contain any clearly labeled pseudocode or algorithm blocks.
Open Source Code No The paper does not provide any statement or link for open-source code related to the described methodology.
Open Datasets Yes In this section, we carry out comprehensive experiments on three benchmark datasets, i.e., Oxford5K (Philbin et al. 2007) , Paris6K (Philbin et al. 2008) and Sculptures6k (Arandjelovi c and Zisserman 2011).
Dataset Splits No The paper mentions using Oxford5K, Paris6K, and Sculptures6k datasets but does not provide specific details on train/validation/test splits (percentages, counts, or explicit splitting methodology).
Hardware Specification No The paper mentions using a '16-layer VGG network' but does not specify any hardware (GPU, CPU, cloud instance type) used for running the experiments.
Software Dependencies No The paper refers to using a 'convolutional neural network (CNN)' and specific architectures like 'VGG network', but does not provide specific software names with version numbers (e.g., TensorFlow, PyTorch, Caffe with versions) for reproducibility.
Experiment Setup Yes Each proposal is resized to the size 448 448 prior to being fed into the network. The 512-dimensional local features are extracted from the last convolutional layer conv5 3. ... the cluster number of VLAD is set to be 64... In the implementation, we fix the number of non-zero elements z in both FOM and SCM as 3 and changes the number of atoms t from 5 to 20. ... the number of proposals n to generate fuzzy objects is set as 300, the number of fuzzy objects t is set as 20 and the number of neighborhoods m in Eq.(8) is set as 20. ... In our implementation, we set s as 20, which is equal to the number of fuzzy objects.