Intelligent and Multimedia Science Laboratory

Visual-based
Geomagnetic-based
Wi-Fi Fingerprint-based
Multi-modal Fusion-based
Fingerprint Database

Introduction

Visual Indoor Localization

	LLM‐Loc: Bootstrap Single-Image Indoor Localization with Large Language Model Qun Niu, Tao Chen, Xing Zhang, Yifan Wang, Ning Liu* intro: Visual indoor localization has gathered significant attention, mainly due to the ubiquity and distinctiveness of visual textures. However, visual texture-based approaches usually suffer from extensive site survey, a lack of generality to different sites and are prone to appearance changes, hindering their practical deployment. We propose LLM-Loc, which combines large language models (LLMs), ubiquitous scene texts and line segments for positioning, as texts and lines present more stable and robust spatial semantics. LLM-Loc comprises two stages: coarse area inference with scene texts and fine positioning with contour lines. To reduce the impact of text diversity and environmental noises on the area identification, we combine the reasoning capability of LLM and semantic priors for area inference. To reduce the impact of occlusion on contour extraction, we propose a collaborative contour refinement algorithm, where we leverage homograhy constraints and refine contour lines using images taken by others. Using the refined contour image, we propose a relative similarity ranking scheme and estimate fine user locations with orientation and map constraints. Extensive experimental results in different trial sites show LLM-Loc is accurate (mean localization errors 1.61m and 1.34m in the food plaza and shopping mall) and reduces positioning error at least by 39%. Expert Systems with Applications (ESWA), 2025 (中科院1区) [论文]
	NeRF‐VLD: Efficient Visual Landmark Database Construction via Scene Constraints Tao Chen, Qun Niu, Ning Liu intro:* We propose a novel method called NeRF-VLD, which first introduces Neural Radiance Fields (NeRF) to build visual landmark databases efficiently. NeRF-VLD generates images from a small number of sources, thus eliminating the need for extensive site surveys and consequently reducing deployment costs significantly. To improve the quality of generated images, we introduce two scene-constrained modules. In the first module, sparse depth maps provided by Structure-from-Motion are introduced as additional geometric constraints. In the second module, scene regularization is applied to image pixels where depth information is unavailable, constraining the volume density of sample points near the camera plane to alleviate ghostly artifacts problem. Experimental results demonstrate that NeRF-VLD can achieve comparable landmark identification and positioning accuracy to real-image databases while reducing the number of image acquisition by 97%. Compared with full manual survey, NeRF-VLD reduces survey and update costs and improves the applicability of vision-based service. 2025 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), 2025 (CCF-B) [Paper]
	Resource efficient and Automated Image-based Indoor Localization Qun Niu, Mingkuan Li, Suining He, Chengying Gao, S.-H. Gary Chan and Xiaonan Luo ACM Transactions on Sensor Networks (TOSN), 2019 (CCF-B)* [Introduction] [Paper]
	SweepLoc: Automatic Video-based Indoor Localization by Camera Sweeping Mingkuan Li, Ning Liu, Qun Niu, Chang Liu, S.-H. Gary Chan, Chengying Gao Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT/UbiComp), 2018 (CCF-A)* [Introduction] [Paper]

Geomagnetic Indoor Localization

	Spatio-temporal Constrained Geomagnetic Indoor Localization with Arbitrary Walking Speed Ning Liu, Gezhi Peng, Hua-Bao Ling, Qun Niu, and Tao He* intro: We propose a speed-adaptive geomagnetic sequence reconstruction algorithm to address the issue of heterogeneity in geomagnetic sequences caused by varying walking speeds along the same path. To tackle the high response time problem in indoor positioning with long geomagnetic sequences, we design a self-attention mechanism to effectively extract temporal features for initial position estimation. Furthermore, we apply an encoder-decoder strategy to leverage spatial contextual information for continuous positioning. IEEE Internet of Things Journal (IoT), 2025 (中科院1区) [Paper]
	GC-Loc: A Graph Attention Based Framework for Collaborative Indoor Localization Using Infrastructure-free Signals Tao He, Qun Niu and Ning Liu Intro: We introduce the collaboration mechanism and propose a graph attention based collaborative indoor localization framework, termed GC-Loc, which provides another perspective for efficient indoor localization. GC-Loc utilizes multiple discrete signal fingerprints collected by several users as input for collaborative localization. Specifically, we first construct an adaptive graph representation to efficiently model the relationships among the collaborative fingerprints. Then taking state-of-the-art GAT model as basic unit, we design a deep network with the residual structure and the hierarchical attention mechanism to extract and aggregate the features from the constructed graph for collaborative localization. Finally, we further employ ensemble learning mechanism in GC-Loc and devise a location refinement strategy based on model consensus for enhancing the robustness of GC-Loc. Interactive, Mobile, Wearable and Ubiquitous Technologies (UbiComp), 2023 (CCF-A) [Paper]
	Efficient Indoor Localization with Multiple Consecutive Geomagnetic Sequences Hui Zhuang, Tao He, Qun Niu and Ning Liu* Intro: We propose an efficient localization approach for continuous positioning scenes (most common in reality), which first utilize short geomagnetic sequences as input, alleviating high response time, and propose an efficient single position estimation model, taking advantage of modified transformer to estimate position for each independent short sequence. Noticing the temporal dependency and the spatial consistency constraint during continuous positioning, we further propose a joint position estimation model to capture the correlations among consecutive short sequences, achieving higher accuracy with multiple short sequences. International Conference on Computer Communications and Networks (ICCCN, 2022) (CCF-C) [Paper]
	MAIL: Multi-Scale Attention-Guided Indoor Localization Using Geomagnetic Sequences Qun Niu, Tao He, Ning Liu, Suining He, Xiaonan Luo, Fan Zhou Intro: We propose MAIL, a multi-scale attention-guided indoor localization network. Our key contributions are as follows. First, instead of extracting a single holistic feature from an input sequence directly, we design a scale-based feature extraction unit that takes variational anomalies at different scales into consideration. Second, we propose an attention generation scheme that identifies attention values for different scales. Rather than setting fixed numbers, MAIL learns them adaptively with the input sequence, thus increasing its adaptability and generality. Third, guided by attention values, we fuse multi-scale features by paying more attention to prominent ones and estimate current location with the fused feature. Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT/UbiComp), 2020 (CCF-A) [Paper]
	Indoor Localization with Spatial and Temporal Representations of Signal Sequences Tao He, Qun Niu, Suining He and Ning Liu IEEE Global Communications Conference (GLOBECOM), 2019 (CCF-C) [Paper]

Wi-Fi Fingerprint-based Indoor Localization

Maxlifd: Joint Maximum Likelihood Localization Fusing Fingerprints and Mutual Distances

Suining He, S.-H. Gary Chan, Lei Yu and Ning Liu*

IEEE Transactions on Mobile Computing (TMC), 2019 (CCF-A)
[Paper]

SLAC: Calibration-Free Pedometer-Fingerprint Fusion for Indoor Localization

Suining He, S.-H. Gary Chan, Lei Yu and Ning Liu*

IEEE Transactions on Mobile Computing (TMC), 2018 (CCF-A)
[Paper]

Multi-modal Fusion-based Indoor Localization

DeepNavi: A Deep Signal-Fusion Framework for Accurate and Applicable Indoor Navigation

Qun Niu, Ning Liu*, Jianjun Huang, Yangze Luo, Suining He, Tao He, S.-H. Chan and Xiaonan Luo

Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT/UbiComp), 2019 (CCF-A)
[Paper]

Incrementive Reconstruction of Fingerprint Database

Fast Radio Map Construction with Domain Disentangled Learning for Wireless Localization

Weina Jiang, Lin Shi, Qun Niu and Ning Liu*

Intro: We propose an accurate and generalizable framework for efficient radio map construction, which takes advantage of readily-available fine-grained radio maps and constructs fine-grained radio maps of a new site with a small proportion of measurements in it. Specifically, we regard radio maps as domains and propose a Radio Map construction approach based on Domain Adaptation (RMDA). We first employ the domain disentanglement feature extractor to learn domain-invariant features for aligning the source domains (available radio maps) with the target domain (initial radio map) in the domain-invariant latent space. Furthermore, we propose a dynamic weighting strategy, which learns the relevancy of the source and target domain in the domain adaptation. Then, we extract the domain-specific features based on the site's floorplan and use them to constrain the super-resolution of the domain-invariant features.

Interactive, Mobile, Wearable and Ubiquitous Technologies (UbiComp), 2023 (CCF-A)
[Paper]

Adaptive Radio Map Reconstruction via Adversarial Wireless Fingerprint Learning

Weina Jiang, Qun Niu, Suining He and Ning Liu*

Intro: We propose a Radio Map Reconstruction framework (RMRec), which adopts adversarial learning to efficiently reconstruct the latest radio map with new signal samples collected at a small portion of Reference Points (RPs). The reconstruction model we built reveals the inherent spatial relations of the Wi-Fi signals in a large-scale building structure and by which the coarse-grained radio map is mapped into the corresponding fine-grained one, thus reducing the cost of the site survey significantly. The adversarial mechanism in RMRec enhances the textural features of the updated radio map, consequently improving the localization service. Meanwhile, we employ the scene-constrained downsample method and the CutPaste data augmentation to improve our model’s reconstruction accuracy and transferability. Besides, we design a non-uniform sampling strategy to reduce the site survey cost by allocating different selection rates for each subarea according to its anti-noise ability for location service.

Neural Computing and Applications (NCA), 2023 (中科院 2区)
[Paper]

RecNet: A Convolutional Network for Efficient Radiomap Reconstruction

Q. Niu, Y. Nie, S. He, N. Liu and X. Luo

IEEE International Conference on Communications (ICC), 2018 (CCF-C)
[Paper]