Robust Place Recognition Using Illumination-compensated Image-based Deep Convolutional Autoencoder Features

Park, Chansoo; Chae, Hee-Won; Song, Jae-Bok

doi:10.1007/s12555-019-0891-x

Cited by 9 publications

(4 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Tis type of neural network can learn efcient image representations in an unsupervised manner, and thus is well suited for VPR tasks that lack high-quality labeled data. It is an appropriate case for a customized CNNbased descriptor, based on its outstanding performance as shown by Gao and Zhang [16] and Park et al [81]. In our work, the encoder of our AE is similar to the implementation of [46] and the decoder is composed of deconvolution and unpooling layers.…”

Section: Customized Cnn-based Descriptorsmentioning

confidence: 99%

Evaluation of Global Descriptor Methods for Appearance-Based Visual Place Recognition

Ma²,

Wang

et al. 2023

Journal of Robotics

View full text Add to dashboard Cite

Visual place recognition (VPR) is considered among the most challenging problems due to the extreme variations in appearance and viewpoint. Essentially, appearance-based VPR can be considered as an image retrieval task, thus the key is to accurately and efficiently describe the images. Recently, global descriptor methods have attracted substantial attention from the VPR community, which has contributed to numerous important outcomes. Despite the growing number of global descriptors presented, little attention has been paid to the comparison and evaluation of these methods and so it remains difficult for researchers to disentangle the factors that led to better performance. This study provided comprehensive insight into global descriptors from a practical application perspective. We present a systematic evaluation that integrates 15 commonly used global descriptors, 6 benchmark datasets, and 5 evaluation metrics, and subsequently extended this evaluation to discuss the key factors impacting the matching performance and computational efficiency. We also report practical suggestions for constructing promising CNN descriptors, based on the experimental conclusions. Our analysis reveals both advantages and limitations of three different types of global descriptors, including handcrafted features-based ones, off-the-shelf CNN-based ones, and customized CNN-based ones. Finally, we evaluate the practicality of reported global descriptors to mediate the trade-offs between matching performance and computational efficiency.

show abstract

Section: Customized Cnn-based Descriptorsmentioning

confidence: 99%

Evaluation of Global Descriptor Methods for Appearance-Based Visual Place Recognition

Ma²,

Wang

et al. 2023

Journal of Robotics

View full text Add to dashboard Cite

show abstract

“…Let ||CVw,t|| be the number of visits of tourist w to destination or attraction in time period t; CSw,t,t΄ be the similarity of the selection of destination or attraction by tourist w between time periods t and t΄. Then, the value of CSw,t,t΄ can be calculated by: (14) The similarity between time periods t and t΄ can be calculated by:…”

Section: Time Periodmentioning

confidence: 99%

“…The advancement of artificial intelligence and data mining has promoted the application of depth convolutional neural network (CNN) in image processing. Many researchers have constructed training samples based on online images of tourist attractions to train neural networks, for the purpose of image matching, identification, and positioning [13][14][15][16]. Lowry and Milford [17] developed a trained deep neural network not limited to a specific task or dataset, and obtained image feature descriptors like sum pooling of convolutions (SPoC), maximum activation of convolutions (MAC), and regional MAC (rMAC) from the response of convolutional layer; the deep neural network has a good generalization ability and high retrieval accuracy.…”

Section: Introductionmentioning

confidence: 99%

Location Identification and Personalized Recommendation of Tourist Attractions Based on Image Processing

Zhang¹,

Liu²,

Liu³

et al. 2021

View full text Add to dashboard Cite

Currently, tourists tend to plan travel routes and itineraries by searching for relevant information on tourist attractions via the Internet and intelligent terminals. However, it is difficult to achieve good retrieval effect on tourist attraction images with text labels. Based on deep learning, the visual location identification faces such defects as frequent mismatching, high probability of weak matching, and long execution time. To solve these defects, this paper puts forward a novel method for location identification and personalized recommendation of tourist attractions based on image processing. Specifically, the authors detailed the ideas and steps of the location identification algorithm for tourist attractions. The algorithm, grounded on hash retrieval, encompasses two stages: an offline stage, and an online stage. Besides, a personalized recommendation model for tourist attractions based on geographical location and time period. Finally, the proposed algorithm and model were proved accurate and effective through experiments.

show abstract

“…Another deep learning structure, the autoencoder, has been also used for visual place recognition because the output of each layer can be used as an image descriptor. Oh and Lee [16] used a deep convolutional autoen-coder (CAE) for feature extraction, and Park [17] proposed an illumination-compensated CAE for robust place recognition.…”

Section: Introductionmentioning

confidence: 99%

Variational Bayesian Approach to Condition-Invariant Feature Extraction for Visual Place Recognition

Eoh

2021

Applied Sciences

View full text Add to dashboard Cite

As mobile robots perform long-term operations in large-scale environments, coping with perceptual changes becomes an important issue recently. This paper introduces a stochastic variational inference and learning architecture that can extract condition-invariant features for visual place recognition in a changing environment. Under the assumption that a latent representation of the variational autoencoder can be divided into condition-invariant and condition-sensitive features, a new structure of the variation autoencoder is proposed and a variational lower bound is derived to train the model. After training the model, condition-invariant features are extracted from test images to calculate the similarity matrix, and the places can be recognized even in severe environmental changes. Experiments were conducted to verify the proposed method, and the experimental results showed that our assumption was reasonable and effective in recognizing places in changing environments.

show abstract

Robust Place Recognition Using Illumination-compensated Image-based Deep Convolutional Autoencoder Features

Cited by 9 publications

References 13 publications

Evaluation of Global Descriptor Methods for Appearance-Based Visual Place Recognition

Evaluation of Global Descriptor Methods for Appearance-Based Visual Place Recognition

Location Identification and Personalized Recommendation of Tourist Attractions Based on Image Processing

Variational Bayesian Approach to Condition-Invariant Feature Extraction for Visual Place Recognition

Contact Info

Product

Resources

About