SeqSLAM: Visual route-based navigation for sunny summer days and stormy winter nights
The DOI matches its Crossref record; 20 of the 22 references it cites were found in a registry or an index.
Abstract
Learning and then recognizing a route, whether travelled during the day or at night, in clear or inclement weather, and in summer or winter is a challenging task for state of the art algorithms in computer vision and robotics. In this paper, we present a new approach to visual navigation under changing conditions dubbed SeqSLAM. Instead of calculating the single location most likely given a current image, our approach calculates the best candidate matching location within every local navigation sequence. Localization is then achieved by recognizing coherent sequences of these “local best matches”. This approach removes the need for global matching performance by the vision front-end - instead it must only pick the best match within any short sequence of images. The approach is applicable over environment changes that render traditional feature-based techniques ineffective. Using two car-mounted camera datasets we demonstrate the effectiveness of the algorithm and compare it to one of the most successful feature-based SLAM algorithms, FAB-MAP. The perceptual change in the datasets is extreme; repeated traverses through environments during the day and then in the middle of the night, at times separated by months or years and in opposite seasons, and in clear weather and extremely heavy rain. While the feature-based method fails, the sequence-based algorithm is able to match trajectory segments at 100% precision with recall rates of up to 60%.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“This has led to develop new methods that explicitly account for such variations by matching sequences [173], gathering different visual appearances into a unified representation [48], or using spatial as well as appearance information [106].”
From Past, Present, and Future of Simultaneous Localization and Mapping: Toward the Robust-Perception Age (2016) · cites this paper as [173]
Citations
Cites
Object recognition from local scale-invariant features
Lowe, D. G. (1999) · Proceedings of the Seventh IEEE International Conference on Computer Vision · doi:10.1109/iccv.1999.790410 · Cited by 16,472 · cited 3×
“Many place recognition techniques rely on feature-finding algorithms such as SIFT [9] and SURF [10] which, despite their impressive rotation and scale invariant properties, are inherently unsuitable when dealing with extreme perceptual M.J. Milford and G.F. Wyeth are with the School of Electrical Engineering and Compu…”
MethodSURF: Speeded Up Robust Features
Bay, H., Tuytelaars, T., & Van Gool, L. (2006) · Lecture Notes in Computer Science · doi:10.1007/11744023_32 · Cited by 14,955
MonoSLAM: Real-time single camera SLAM
Davison, A. J., Reid, I. D., Molton, N. D., & Stasse, O. (2007) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2007.1049 · Cited by 3,959 · cited 1×
“isual recognition of places and place sequences in the field of robotic navigation has advanced rapidly in recent years, with several techniques [1-5] now being competitive with laser range-based techniques.”
BackgroundA solution to the simultaneous localization and map building (SLAM) problem
Dissanayake, M. W. M. G., Newman, P., Clark, S., Durrant-Whyte, H. F., & Csorba, M. (2001) · IEEE Transactions on Robotics and Automation · doi:10.1109/70.938381 · Cited by 2,661
Matching Local Self-Similarities across Images and Videos
Shechtman, E., & Irani, M. (2007) · 2007 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2007.383198 · Cited by 1,076 · cited 1×
“One promising step [15] towards reducing the reliance on consistent feature detection uses the self-similarity of objects within a frame or video sequence to robustly recognize objects under varying conditions.”
BackgroundIncremental mapping of large cyclic environments
Gutmann, J.-S., & Konolige, K. (2003) · Proceedings 1999 IEEE International Symposium on Computational Intelligence in Robotics and Automation. CIRA'99 (Cat. No.99EX375) · doi:10.1109/cira.1999.810068 · Cited by 568 · cited 1×
“There have been attempts to increase the robustness of visual techniques to environment change by performing sensor fusion with lasers at the level of landmarks [13] or dense depth maps [14].”
BackgroundFrameSLAM: From Bundle Adjustment to Real-Time Visual Mapping
Konolige, K., & Agrawal, M. (2008) · IEEE Transactions on Robotics · doi:10.1109/tro.2008.2004832 · Cited by 532
RatSLAM: a hippocampal model for simultaneous localization and mapping
Milford, M. J., Wyeth, G. F., & Prasser, D. (2004) · IEEE International Conference on Robotics and Automation, 2004. Proceedings. ICRA '04. 2004 · doi:10.1109/robot.2004.1307183 · Cited by 399 · cited 6×
“The approach taken in this work is partly inspired by the author's previous work in vision-based mapping in RatSLAM [16] [17] [18].”
MethodInfluentialSelf-citationMapping a Suburb With a Single Camera Using a Biologically Inspired SLAM System
Milford, M. J., & Wyeth, G. F. (2008) · IEEE Transactions on Robotics · doi:10.1109/tro.2008.2004520 · Cited by 378 · cited 7×
“The approach taken in this work is partly inspired by the author's previous work in vision-based mapping in RatSLAM [16] [17] [18].”
BackgroundMethodInfluentialSelf-citationDynamic Maps for Long-Term Operation of Mobile Service Robots
Biber, P., & Duckett, T. (2005) · Robotics: Science and Systems I · doi:10.15607/rss.2005.i.003 · Cited by 334 · cited 1×
“The problem of visual place recognition across time has received far less attention, with relatively few attempts to address the problem [6, 7], including high dynamic range approaches [8].”
Background
Cited by
Items 301–400 of 987
Erasing bad memories: Agent-side summarization for long-term mapping
Dymczyk, M., Schneider, T., Gilitschenski, I., Siegwart, R., & Stumm, E. S. (2016) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2016.7759673 · Cited by 26
“Current state-of-the-art place recognition approaches range from using holistic image retrieval [12], over features obtained using deep-learning techniques [13], to local point features [14].”
BackgroundDeepSeqSLAM: A Trainable CNN+RNN for Joint Global Description and Sequence-based Place RecognitionPreprint
Chancán, M., & Milford, M. J. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2011.08518 · Cited by 25
“In contrast to classical sequence-based methods [30, 35, 18, 17]—which often require velocity information or reference-query image sequences being of the same size—and recent neuro-inspired systems [39, 33, 11], our approach does not require velocity data and can work with query traversals of any size—only requiring p…”
MethodResultInfluentialSelf-citationA compact neuromorphic system for ultra–energy-efficient, on-device robot localization
Hines, A., Milford, M. J., & Fischer, T. (2025) · Science Robotics · doi:10.1126/scirobotics.ads3968 · Cited by 22
“Online place matching Place matching was performed using real-time sequence matching ( 46, 71 ) to enhance the precision of our VPR model on the Speck ™ .”
Self-citationDOSeqSLAM: Dynamic On-line Sequence Based Loop Closure Detection Algorithm for SLAM
Tsintotas, K. A., Bampis, L., & Gasteratos, A. C. (2018) · IEEE International Conference on Imaging Systems and Techniques (IST) · doi:10.1109/ist.2018.8577113 · Cited by 25
“SeqSLAM [8], one of the most renown algorithms in sequence-based visual place recognition [9], [10], [11], [12], [13], tackles the recognition task in an off-line procedure by comparing the query sequence against the database groups-of-images for the appropriate sequence match selection.”
MethodAppearance-Based Loop Closure Detection with Scale-Restrictive Visual Features
Tsintotas, K. A., Giannis, P., Bampis, L., & Gasteratos, A. C. (2019) · Lecture notes in computer science · doi:10.1007/978-3-030-34995-0_7 · Cited by 22
Garg, S., & Milford, M. J. (2020) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra40945.2020.9196827 · Cited by 22
“The topic has been extensively researched for decades [1] with researchers exploring different aspects of the problem, such as dealing with appearance [2], [3] and viewpoint variations [4], [5], and large-scale localization [6], [7], [8] and navigation [9].”
BackgroundMethodResultInfluentialSelf-citationPerceptual Aliasing++: Adversarial Attack for Visual SLAM Front-End and Back-End
Ikram, M. H., Khaliq, S., Anjum, M. L., & Hussain, W. (2022) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2022.3150031 · Cited by 26
Dataset and Benchmark: Novel Sensors for Autonomous Vehicle Perception
Carmichael, S., Buchan, A. D., Ramanagopal, M. S., Ravi, R., Vasudevan, R., & Skinner, K. A. (2024) · The International Journal of Robotics Research · doi:10.1177/02783649241273554 · Cited by 32
“We use a 10 meter localization radius (denoted with r = 10 m) to determine correct matches for the sequences driven in the same direction as the reference database and, following Garg et al. (2018), we use an 80 meter localization radius (denoted with r = 80 m) for the opposing viewpoint sequences.”
Applications of Spiking Neural Networks in Visual Place Recognition
Hussaini, S., Milford, M. J., & Fischer, T. (2024) · IEEE Transactions on Robotics · doi:10.1109/tro.2024.3508053 · Cited by 29
“Ourcontributionshighlighttheviabilityof SNNsforVPR,offeringscalableandrobustsolutions,andpavingthewayfortheirapplicationinvariousenergy-sensitiverobotic tasks.”
InfluentialSelf-citationTowards persistent visual navigation using SMART
Pepperell, E., Corke, P., & Milford, M. J. (2013) · QUT ePrints (Queensland University of Technology) · Cited by 22
Self-citationDynamic scene models for incremental, long-term, appearance-based localisation
Johns, E., & Yang, G. (2013) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2013.6630953 · Cited by 25
“Note that we do not consider odometry or filtering as in [6], [11], [5] and concentrate solely on the computer vision problem of matching a query image to a location in a database.”
BackgroundTopological map construction and scene recognition for vehicle localization
Lin, H., Yao, C.-W., Cheng, K.-S., & Tran, V. L. (2017) · Autonomous Robots · doi:10.1007/s10514-017-9638-9 · Cited by 22
“Milford and Wyeth (2012) develop a feature-based image retrieval technique called sequential simultaneous localization and mapping (SeqS-LAM), which is able to adapt to the changing environment conditions.”
BackgroundAdapting Semantic Segmentation Models for Changes in Illumination and Camera Perspective
Zhou, W., Zyner, A., Worrall, S., & Nebot, E. (2019) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2019.2891027 · Cited by 23
“In addition to the local data, we also employ images collected from Brisbane Australia [20], [21] and images used for localization [22] to qualitatively cross-test the semantic segmentation models.”
MethodTB-Places: A Data Set for Visual Place Recognition in Garden Environments
Leyva-Vallina, M., Strisciuglio, N., López-Antequera, M., Tyleček, R., Blaich, M., & Petkov, N. (2019) · IEEE Access · doi:10.1109/access.2019.2910150 · Cited by 26
“The second group contains methods based on global image descriptors, such as SeqSLAM (Milford and Wyeth (2012)), which performs place recognition by matching image sequences, instead of single images, and addressing strong illumination (day vs night) and weather (clear vs rain) variations in urban…”
BackgroundMethodInfluentialCompressed Holistic ConvNet Representations for Detecting Loop Closures in Dynamic Environments
Wang, S., Lv, X., Liu, X., & Ye, D. (2020) · IEEE Access · doi:10.1109/access.2020.2982228 · Cited by 22
“SeqSLAM [11] was another successful global descriptionbased loop closure detection system.”
MethodNetHALOC : A learned global image descriptor for loop closing in underwater visual SLAM
Font, F. B., & Burguera, A. (2020) · Expert Systems · doi:10.1111/exsy.12635 · Cited by 22
“Other methods try to overcome certain incapacity of BoW formulas to handle changes on the illumination conditions, joining different appearances into a single representation (Milford & Wyeth, 2012).”
MethodAlijani, F., & Rahtu, E. (2020) · IEEE International Workshop on Multimedia Signal Processing (MMSP) · doi:10.1109/mmsp48831.2020.9287063 · Cited by 21
“Before the rise and dominance of CNNs, VPR methods utilized conventional hand-crafted features such as local and global image descriptors, to match query images with database [16]–[18].”
MethodSeqSLAM++: View-based robot localization and navigation
Oishi, S., Inoue, Y., Miura, J., & Tanaka, S. (2018) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2018.10.014 · Cited by 25
“Among various approaches to resolving the issue, we adopt the SeqSLAMmethod [10], which achieves a high robustness even under extreme view changes by using image sequence matching.”
InfluentialLi, Z., & Zhou, A. (2021) · Journal of Intelligent & Robotic Systems · doi:10.1007/s10846-021-01421-2 · Cited by 21
“SeqSLAM [36], and but also deep learning feature methods, e.”
MethodVISUALIZING ROAD DAMAGE BY MONITORING SYSTEM IN CLOUD
Deepa, D., Vignesh, R., Sivasangari, A. A., Mana, S. C., Samhitha, B. K., & Jose, J. (2020) · INTERNATIONAL JOURNAL OF ELECTRICAL ENGINEERING & TECHNOLOGY · doi:10.34218/ijeet.11.4.2020.022 · Cited by 21
Brain-inspired sensor fusion for navigating robots
Milford, M. J., & Jacobson, A. (2013) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2013.6630980 · Cited by 21
Self-citationAre you ABLE to perform a life-long visual topological localization?
Arroyo, R., Alcantarilla, P. F., Bergasa, L. M., & Romera, E. (2017) · Autonomous Robots · doi:10.1007/s10514-017-9664-7 · Cited by 23
“– Sequences of images are used instead of single images in all the cases with the aim of carrying out a better recognition of places in long-term scenarios, as introduced in some state-of-the-art works such as Milford and Wyeth (2012).”
MethodTomită, M.-A., Zaffar, M., Milford, M. J., McDonald-Maier, K. D., & Ehsan, S. (2021) · IEEE Access · doi:10.1109/access.2021.3107778 · Cited by 21
“The authors of SeqSLAM [11] decided to compare sequences of camera frames instead of single image comparison, thus achieving increased performance in VPR compared to traditional feature-based techniques, when the place is subject to drastic changes.”
BackgroundMethodInfluentialSelf-citationLoopNet: Where to Focus? Detecting Loop Closures in Dynamic Scenes
Osman, H. A., Darwish, N. M., & Bayoumi, A. (2022) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2022.3142901 · Cited by 21
Ren, Z., & Wang, L. (2022) · Sensors · doi:10.3390/s22041463 · Cited by 21
What makes visual place recognition easy or hard?Preprint
Schubert, S., & Neubert, P. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2106.12671 · Cited by 26
“[4], [6]) or the standardization of descriptors [3].”
BackgroundPlaceNet: A multi-scale semantic-aware model for visual loop closure detection
Osman, H. A., Darwish, N. M., & Bayoumi, A. (2022) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2022.105797 · Cited by 21
Arshad, S., & Kim, G.-W. (2023) · International Journal of Control Automation and Systems · doi:10.1007/s12555-021-0927-x · Cited by 20
“SeqSLAM’s poor performance on other datasets is likely due to its assumption of viewpoint invariance, which is violated in these cases.”
MethodInfluentialEv-ReconNet: Visual Place Recognition Using Event Camera With Spiking Neural Networks
Lee, H., & Hwang, H. (2023) · IEEE Sensors Journal · doi:10.1109/jsen.2023.3298828 · Cited by 23
“F. Wyeth, “SeqSLAM: Visual route-based navigation for sunny summer days and stormy winter nights,” in Proc.”
Deep Homography Estimation for Visual Place Recognition
Lu, F., Dong, S., Zhang, L., Liu, B., Lan, X., Jiang, D., & Yuan, C. (2024) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v38i9.28901 · Cited by 20
Leveraging image-based prior in cross-season place recognition
Masatoshi, A., Yuuto, C., Kanji, T., & Kentaro, Y. (2015) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2015.7139961 · Cited by 25
“As suggested by our experimental results, previous research on cross-season VPC have shown that the BoW scene model is not sufficiently discriminative and often fails to capture the appearance changes across seasons [2].”
BackgroundMethodResultInfluentialEnabling Mobile and Wireless Technologies for Smart Cities
Ahmed, E., Imran, M. A., Guizani, M., Rayes, A., Lloret, J., Han, G., & Guibène, W. (2017) · IEEE Communications Magazine · doi:10.1109/mcom.2017.7823341 · Cited by 20
High order visual words for structure-aware and viewpoint-invariant loop closure detection
Bampis, L., Amanatiadis, A. A., & Gasteratos, A. C. (2017) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2017.8206289 · Cited by 20
“Instead, methods like the ones presented by Milford and Wyeth [24] or Arroyo et al. [25] concluded into a global image description, sacrificing some of the viewpoint invariance, to gain robustness over the potential appearance changes.”
MethodReal-Time Visual Place Recognition Based on Analyzing Distribution of Multi-scale CNN Landmarks
Xin, Z., Cui, X., Zhang, J., Yang, Y., & Wang, Y. (2018) · Journal of Intelligent & Robotic Systems · doi:10.1007/s10846-018-0804-x · Cited by 20
“It has been verified that holistic descriptors achieve superior performance in changing environments against local descriptors [6].”
BackgroundMethodOmnidirectional Multisensory Perception Fusion for Long-Term Place Recognition
Siva, S., & Zhang, H. (2018) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2018.8461042 · Cited by 20
“When environments show changes in illumination conditions, global features outperform local features [1], [3], [9].”
BackgroundMethodInfluentialHigh-fidelity simulation for evaluating robotic vision performance
Skinner, J., Garg, S., Sünderhauf, N., Corke, P., Upcroft, B., & Milford, M. J. (2016) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2016.7759425 · Cited by 26
“On the other hand, use of global image descriptors like BRIEFGIST [22] or patch-normalised downsampled images as in SeqSLAM [14] allows matching across change in conditions,”
MethodResultInfluentialSelf-citationSequence-based mapping for probabilistic visual loop-closure detection
Tsintotas, K. A., Bampis, L., An, S., Fragulis, G. F., Mouroutsos, S. G., & Gasteratos, A. C. (2021) · IEEE International Conference on Imaging Systems and Techniques (IST) · doi:10.1109/ist50367.2021.9651458 · Cited by 19
PlaceNav: Topological Navigation through Place Recognition
Suomela, L., Kalliola, J., Edelman, H., & Kämäräinen, J. (2024) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra57147.2024.10610575 · Cited by 29
“Early approaches [29–32] utilized feature similarity matrices to find the best-matching image sequences.”
MethodDistance metric learning for feature-agnostic place recognition
Chen, Z., Lowry, S. M., Jacobson, A., Ge, Z., & Milford, M. J. (2015) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2015.7353725 · Cited by 19
“presented SeqSLAM [12] which performs localization against image sequences rather than individual images.”
BackgroundMethodInfluentialSelf-citationRobust UAV Visual Teach and Repeat Using Only Sparse Semantic Object Features
Toudeshki, A. G., Shamshirdar, F., & Vaughan, R. (2018) · Canadian Conference on Computer and Robot Vision · doi:10.1109/crv.2018.00034 · Cited by 19
“This problem is also present in methods where low-resolution whole-images are used as decriptors, so we borrow the SeqSLAM technique developed for that domain [5] where localization is done over a contiguous sequence of images.”
MethodInfluentialLearning to Fuse Multiscale Features for Visual Place Recognition
Mao, J., Hu, X., He, X., Zhang, L., Wu, L., & Milford, M. J. (2018) · IEEE Access · doi:10.1109/access.2018.2889030 · Cited by 20
“• ‘‘SeqSLAM’’ [6]: A visual place recognition approach utilizing sequence images.”
BackgroundSelf-citationConvNet and LSH-Based Visual Localization Using Localized Sequence Matching
Qiao, Y., Cappelle, C., Ruichek, Y., & Yang, T. (2019) · Sensors · doi:10.3390/s19112439 · Cited by 20
“Table 3 shows F1 scores obtained for different layers and other state-of-the-art methods like SeqSLAM and FAB-MAP.”
BackgroundMethodInfluentialFilter Early, Match Late: Improving Network-Based Visual Place Recognition
Hausler, S., Jacobson, A., & Milford, M. J. (2019) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros40897.2019.8967783 · Cited by 19
“The place recognition quality score is calculated using the method originally proposed in SeqSLAM [25], where the quality score is the ratio between the score at the best matching template and the next largest score outside a window around the best matching template.”
MethodSelf-citationLookUP: Vision-Only Real-Time Precise Underground Localisation for Autonomous Mining Vehicles
Zeng, F., Jacobson, A., Smith, D., Boswell, N., Peynot, T., & Milford, M. J. (2019) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2019.8794453 · Cited by 19
“SeqSLAM [12] and many other SLAM frameworks [13], [14], [15] are less sensitive but require external sources like GPS or wheel odometry to provide metric information.”
BackgroundSelf-citationFast Sequence-Matching Enhanced Viewpoint-Invariant 3-D Place Recognition
Yin, P., Wang, F., Egorov, A., Hou, J., Jia, Z., & Han, J. (2021) · IEEE Transactions on Industrial Electronics · doi:10.1109/tie.2021.3057025 · Cited by 25
“Compared with the original burst-force sequence-matching method SeqSLAM [11], deeper-resolution-based coarse-to-fine .”
MethodInfluentialArcanjo, B., Ferrarini, B., Milford, M. J., McDonald-Maier, K. D., & Ehsan, S. (2022) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2022.3140827 · Cited by 32
Self-citationFast and Robust Loop-Closure Detection via Convolutional Auto-Encoder and Motion Consensus
Ma, J., Wang, S., Zhang, K., He, Z., Huang, J., & Mei, X. (2021) · IEEE Transactions on Industrial Informatics · doi:10.1109/tii.2021.3120141 · Cited by 21
Patch-NetVLAD+: Learned patch descriptor and weighted matching strategy for place recognition
Cai, Y., Zhao, J., Cui, J., Zhang, F., Feng, T., & Ye, C. (2022) · IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI) · doi:10.1109/mfi55806.2022.9913860 · Cited by 18
“The methods to optimize the matching strategy are mainly based on intra-set similarities [19, 22] and sequence searching [5, 17].”
BackgroundMethodSTUN: Self-Teaching Uncertainty Estimation for Place Recognition
Cai, K., Lu, C. X., & Huang, X. (2022) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros47612.2022.9981546 · Cited by 23
Multi-scale bio-inspired place recognition
Chen, Z., Jacobson, A., Erdem, U. M., Hasselmo, M. E., & Milford, M. J. (2014) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2014.6907109 · Cited by 18
“Some of the most significant vision-based mapping and localization algorithms proposed over the past decade include FAB-MAP [10] [11], MonoSLAM [12], RatSLAM [13] [14] [15] [16], FrameSLAM [17] and SeqSLAM [18] [19].”
MethodSelf-citationLowry, S. M., Wyeth, G. F., & Milford, M. J. (2014) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2014.6906932 · Cited by 18
“To increase robustness over perceptual change, the whole-image visual input is generally combined with other techniques such as range sensing [13] or sequence filtering [11].”
MethodInfluentialSelf-citationInsect navigation: do ants live in the now?
Graham, P., & Mangan, M. (2015) · Journal of Experimental Biology · doi:10.1242/jeb.065409 · Cited by 20
Arroyo, R., Alcantarilla, P. F., Bergasa, L. M., & Romera, E. (2016) · IEEE International Conference on Intelligent Transportation Systems (ITSC) · doi:10.1109/itsc.2016.7795672 · Cited by 18
“A wide set of satisfactory results has validated its performance compared to other open toolboxes in the state of the art, such as OpenFABMAP or OpenSeqSLAM.”
BackgroundInfluentialCheckout my map: Version control for fleetwide visual localisation
Gadd, M., & Newman, P. M. (2016) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2016.7759843 · Cited by 18
“More recent work has involved searching for consistent sequences of image matches [4], which Pepperell et al.”
BackgroundHou, Y., Zhang, H., Zhou, S., & Zou, H. (2017) · IEEE Access · doi:10.1109/access.2017.2698524 · Cited by 19
“As a result, this method has demonstrated the state-of-the-art performance, superior to traditional methods such as FAB-MAP [3] and SeqSLAM [8] under conditions of significant viewpoint and environmental changes [16].”
BackgroundMethodInfluentialOnline semantic mapping of logistic environments using RGB-D cameras
Himstedt, M., & Maehle, E. (2017) · International Journal of Advanced Robotic Systems · doi:10.1177/1729881417720781 · Cited by 18
“Each distribution pðmjÞ is estimated by those points of O that end inside mj according to equation (22).”
BackgroundBampis, L., & Gasteratos, A. C. (2019) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2019.01.004 · Cited by 18
Ferrarini, B., Waheed, M., Waheed, S., Ehsan, S., Milford, M. J., & McDonald-Maier, K. D. (2019) · NASA/ESA Conference on Adaptive Hardware and Systems (AHS) · doi:10.1109/ahs.2019.00011 · Cited by 18
“State-ot-the-art VPR methods are only robust to small viewpoint changes [7], [8] or their computation intensive nature [9] makes their use prohibitive for small UAVs Some recent works [2], [10] proposed VPR pipelines based on local image features as they can be extracted efficiently on resource-constrained hardware.”
BackgroundSelf-citationA Novel Loop Closure Detection Method Using Line Features
Dong, R., Wei, Z., Liu, C., & Kan, J. (2019) · IEEE Access · doi:10.1109/access.2019.2934521 · Cited by 18
“SeqSLAM [32], SeqSLAM2.”
A Novel Approach to Image-Sequence-Based Mobile Robot Place Recognition
Yuan, J., Zhu, W., Dong, X., Sun, F., Zhang, X., Sun, Q., & Huang, Y. (2019) · IEEE Transactions on Systems Man and Cybernetics Systems · doi:10.1109/tsmc.2019.2956321 · Cited by 18
“It is worthy noting that, according to [36], the image features of these two algorithms are the pixel values of drastically downsam-pled incoming images.”
BackgroundGiubilato, R., Gentil, C. L., Vayugundla, M., Schuster, M. J., Vidal‐Calleja, T., & Triebel, R. (2022) · Field Robotics · doi:10.55417/fr.2022053 · Cited by 19
“A different path towards visual place recognition belongs to the SeqSLAM approach [Milford and Wyeth, 2012, Siam and Zhang, 2017], proven to work with some adaptations also on very challenging unstructured and self-similar scenarios [Grixa et al., 2018].”
MethodMultidimensional Particle Filter for Long-Term Visual Teach and Repeat in Changing Environments
Rozsypálek, Z., Rouček, T., Vintr, T., & Krajník, T. (2023) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2023.3244418 · Cited by 18
“Using the robots in uncontrolled and natural environments requires updating the maps continuously [17], [18], knowing which parts of the scene are stable [19], [20], [21], [22] or trying to predict the changes [23], [24], [25].”
BackgroundMethodMRS-VPR: a multi-resolution sampling based global visual place recognition method
Yin, P., He, Y., Srivatsan, R. A., Chen, Y., Li, X.-Q., Zhang, H., Xu, L., Li, L., Jia, Z., & Ji, J. (2019) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2019.8793853 · Cited by 21
“proposed a sequence matching method, SeqSLAM [2].”
MethodLearning Sequence Descriptor Based on Spatio-Temporal Attention for Visual Place Recognition
Zhao, J., Feng-lin, Z., Cai, Y., Tian, G., Mu, W., Ye, C., & Feng, T. (2024) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2024.3354627 · Cited by 20
“SeqSLAM [8] is a pioneering example of sequence matching.”
BackgroundTrajectory-Based Place-Recognition for Efficient Large Scale Localization
Lynen, S., Bosse, M. C., & Siegwart, R. (2016) · International Journal of Computer Vision · doi:10.1007/s11263-016-0947-9 · Cited by 23
“Similarly the work by Milford and Wyeth (2012) remove any discretization of the trajectory into places, by performing correlation-based matching on entire sequences of downsampled images, instead of looking at individual images or descriptors.”
MethodStabilize an Unsupervised Feature Learning for LiDAR-based Place Recognition
Yin, P., Xu, L., Liu, Z., Li, L., Salman, H., He, Y., Xu, W., Wang, H., & Choset, H. (2018) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2018.8593562 · Cited by 17
Robust Photogeometric Localization Over Time for Map-Centric Loop Closure
Park, C., Kim, S., Moghadam, P., Guo, J., Sridharan, S., & Fookes, C. (2019) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2019.2895262 · Cited by 19
Chen, L., Jin, S., & Xia, Z. (2021) · Sensors · doi:10.3390/s21010310 · Cited by 19
“SeqSLAM [29] is a visual place recognition technique using confusion matrix by subtracting patch-normalized sequences of images to find the matched place, which shows robust to seasonal and illumination variations.”
MethodCrowdDriven: A New Challenging Dataset for Outdoor Visual Localization
Jafarzadeh, A., López-Antequera, M., Gargallo, P., Kuang, Y., Toft, C., Kahl, F., & Sattler, T. (2021) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv48922.2021.00970 · Cited by 22
“While localization approaches use metric maps, visual place recognition methods typically rely on topological scene representations [19, 20, 27, 28, 49, 50, 78], and thus do not directly provide a camera pose estimate.”
BackgroundChen, T., Pu, F., Chen, H., & Liu, Z. (2022) · Remote Sensing · doi:10.3390/rs14092033 · Cited by 18
“Datasets related to visual-inertial odometry (VIO), simultaneous localization and mapping (SLAM), and autonomous driving released by various research institutions and colleges in the past decade have greatly promoted the development of these technologies, e.g., UTIAS Multi-Robot [1], San Francisco Landmark [2], SeqSLA…”
BackgroundAdaptive-Attentive Geolocalization From Few Queries: A Hybrid Approach
Paolicelli, V., Berton, G., Montagna, F., Masone, C., & Barbara Caputo, B. (2022) · Frontiers in Computer Science · doi:10.3389/fcomp.2022.841817 · Cited by 21
“, 2020), autonomous robots and vehicles (Cummins and Newman, 2008; Milford and Wyeth, 2012; McManus et al., 2014; Tomita et al., 2021), the cataloging of archival imagery (Aubry et al.”
Semantic Map Building Based on Object Detection for Indoor Navigation
Yang, J., Zhang, J., Wang, G., & Li, M. (2015) · International Journal of Advanced Robotic Systems · doi:10.5772/62040 · Cited by 17
Williams, H. A., Browne, W. N., & Carnegie, D. A. (2016) · Applied Soft Computing · doi:10.1016/j.asoc.2016.11.036 · Cited by 17
Manifold Regularization Graph Structure Auto-Encoder to Detect Loop Closure for Visual SLAM
Wang, Z., Peng, Z., Guan, Y., & Wu, L. (2019) · IEEE Access · doi:10.1109/access.2019.2914943 · Cited by 17
High Accuracy and Low Complexity Lidar Place Recognition Using Unitary Invariant Frobenius Norm
Wang, W., Min, H., Wu, X., Hou, X., Li, Y., & Zhao, X. (2022) · IEEE Sensors Journal · doi:10.1109/jsen.2022.3221979 · Cited by 16
“NetVLAD [28] uses a fixed-size vector and SeqSLAM uses a reduced image [29] to represent the original image.”
MethodMatching-range-constrained real-time loop closure detection with CNNs features
Bai, D., Wang, C., Zhang, B., Yi, X., & Tang, Y. (2016) · Robotics and Biomimetics · doi:10.1186/s40638-016-0047-x · Cited by 16
“Methods that address the LCD problem span from matching sequences of images [11], [12], transforming images”
BackgroundVision-based Markov localization for long-term autonomy
Naseer, T., Suger, B., Ruhnke, M., & Burgard, W. (2016) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2016.11.008 · Cited by 16
Change Detection with Global Viewpoint Localization
Murase, T., Kanji, T., & Takayama, A. (2017) · Asian Conference on Pattern Recognition · doi:10.1109/acpr.2017.21 · Cited by 16
Deep Supervised Hashing with Similar Hierarchy for Place Recognition
Wu, L., & Wu, Y. (2019) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros40897.2019.8968599 · Cited by 16
“Instead of calculating the most similar location from a given single image, SeqSLAM [10, 11] selects the best candidate to match place from image sequence.”
BackgroundMethodResultInfluentialIntelligent Reference Curation for Visual Place Recognition Via Bayesian Selective Fusion
Molloy, T. L., Fischer, T., Milford, M. J., & Nair, G. N. (2020) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2020.3047791 · Cited by 16
“8 shows that the performance of our approach on the randomized query from our previous study is improved via integration with an existing sequence matching approach (SeqSLAM [5]) with a sequence length of 5.”
MethodSelf-citationFast and Memory Efficient Graph Optimization via ICM for Visual Place Recognition
Schubert, S., Neubert, P., & Protzel, P. (2021) · Robotics: Science and Systems · doi:10.15607/rss.2021.xvii.091 · Cited by 16
“Established approaches are the standardization of descriptors [31] and sequence processing (e.g., [22, 25]).”
MethodVisual Loop-Closure Detection via Prominent Feature Tracking
Papapetros, I. T., Balaska, V. E., & Gasteratos, A. C. (2022) · Journal of Intelligent & Robotic Systems · doi:10.1007/s10846-022-01581-9 · Cited by 16
“[26] proposed a local normalization scheme, allowing for locations with fainter similarity to be assessed as a possible match, whilst the final filtering was achieved based on the high similarity trail angle each location endows in the confusion matrix [44, 45].”
BackgroundPredicting to Improve: Integrity Measures for Assessing Visual Localization Performance
Carson, H., Ford, J., & Milford, M. J. (2022) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2022.3191205 · Cited by 19
Self-citationAn efficient visual loop closure detection method in a map of 20 million key locations
Wu, J., Zhang, H., & Guan, Y. (2014) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2014.6906955 · Cited by 16
“As a result, the complexity of this verification step is constant and independent of the map size.”
BackgroundPredicting the change - A step towards life-long operation in everyday environments
Suenderhauf, N., Neubert, P., & Protzel, P. (2013) · QUT ePrints (Queensland University of Technology) · Cited by 16
Kuse, M., & Shen, S. (2021) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2021.103813 · Cited by 17
Featureless Visual Processing for SLAM in Changing Outdoor Environments
Milford, M. J., & George, A. (2013) · Springer tracts in advanced robotics · doi:10.1007/978-3-642-40686-7_38 · Cited by 16
“The work presented here builds on previous research including mapping of a suburban road network at different times of day [15, 16] and sequence-based localization on road networks [17, 18].”
BackgroundMethodSelf-citationLarge-scale 3D outdoor mapping and on-line localization using 3D-2D matching
Sakai, T., Koide, K., Miura, J., & Oishi, S. (2017) · IEEE/SICE International Symposium on System Integration (SII) · doi:10.1109/sii.2017.8279325 · Cited by 16
Three-Dimensional Dense Reconstruction: A Review of Algorithms and Datasets
Lee, Y. (2024) · Sensors · doi:10.3390/s24185861 · Cited by 20
Enhancing Visual Place Recognition via Fast and Slow Adaptive Biasing in Event Cameras
Nair, G. B., Milford, M. J., & Fischer, T. (2024) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros58592.2024.10802384 · Cited by 15
“Pseudo ground truth robot positions in a fixed world frame are obtained using the SLAM Toolbox ROS package [39], which runs a full SLAM pipeline that takes the raw odometry from the wheel-encoders and LiDAR data as input to provide accurate ground truth position information in a global coordinate…”
Self-citationLong exposure localization in darkness using consumer cameras
Milford, M. J., Turner, I., & Corke, P. (2013) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2013.6631105 · Cited by 15
“The work in this paper extends on recent work [4] that introduced the SeqSLAM algorithm and set the benchmark for passive, conventional camera-based localization across day-night cycles.”
BackgroundMethodSelf-citationVisual Localization across Seasons Using Sequence Matching Based on Multi-Feature Combination
Qiao, Y., Cappelle, C., & Ruichek, Y. (2017) · Sensors · doi:10.3390/s17112442 · Cited by 15
“Recently, several approaches such as SeqSLAM have proved that recognizing places through sequences of images is more robust and effective [7].”
BackgroundMethodInfluentialSiamese-ResNet: Implementing Loop Closure Detection based on Siamese Network
Qiu, K., Ai, Y., Tian, B., Wang, B., & Cao, D. (2018) · IEEE Intelligent Vehicles Symposium · doi:10.1109/ivs.2018.8500465 · Cited by 15
“SeqSLAM [31] was proposed by Michael J.”
Loop closure detection using CNN words
Liu, Q., & Duan, F. (2019) · Intelligent Service Robotics · doi:10.1007/s11370-019-00284-9 · Cited by 15
“In this section, we perform experiments to evaluate our LCD system.”
BackgroundMethodInfluentialLong-Term Knowledge Distillation of Visual Place Classifiers
Tomoe, H., & Kanji, T. (2019) · IEEE International Conference on Intelligent Transportation Systems (ITSC) · doi:10.1109/itsc.2019.8917199 · Cited by 18
“The function Cs,s′ (s′ < s) splits the students into the groups {1, · · · ,s′}, and {s′+ 1, · · · ,s}, for the former group, it assigns previous and/or current teachers with the number of teachers in range [1,5], while for the latter group, it only assigns the current season’s teacher (i.”
BackgroundProbabilistic Appearance-Invariant Topometric Localization With New Place Awareness
Xu, M., Fischer, T., Sünderhauf, N., & Milford, M. J. (2021) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2021.3096745 · Cited by 19
“The VPR problem is commonly treated as a template matching problem [20]; we assume the map consists of a set of reference “places,” and each place has an associated template extracted from either single images [1], [4], [21] or image sequences [22]–[24].”
BackgroundSelf-citationCollision‐free local planner for unknown subterranean navigation
Jung, S., Lee, H., Shim, D. H., & Agha–mohammadi, A. (2021) · ETRI Journal · doi:10.4218/etrij.2021-0087 · Cited by 15
“Unavailability of GPS signals and limited lighting conditions render the perception problem extremely difficult [3].”
Background3D Recognition Based on Sensor Modalities for Robotic Systems: A Survey
Manzoor, S., Joo, S.-H., Kim, E., Bae, S.-H., In, G.-G., Pyo, J.-W., & Kuc, T. (2021) · Sensors · doi:10.3390/s21217120 · Cited by 15
Hierarchical Loop Closure Detection for Long-term Visual SLAM with Semantic-Geometric Descriptors
Singh, G., Wu, M., Lam, S.-K., & Minh, D. V. (2021) · IEEE International Conference on Intelligent Transportation Systems (ITSC) · doi:10.1109/itsc48978.2021.9564866 · Cited by 17
“SeqSLAM [12] uses sequence of images to improve the loop detection accuracy. iBOW-LCD learns the vocabulary incrementally and therefore doesn’t require any training stage.”
MethodFHT-Map: Feature-Based Hybrid Topological Map for Relocalization and Path Planning
Song, K., Liu, W., Chen, G., Xu, X., & Xiong, Z. (2024) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2024.3392493 · Cited by 17
Metadata
- Authors
- Milford, M. J., & Wyeth, G. F.
- Affiliations
- Queensland University of Technology (AU)
- Venue
- 2012 IEEE International Conference on Robotics and Automation
- Publisher
- IEEE
- Published
- 2012-05
- Pages
- 1643-1649
- ISSN
- 1050-4729 (ISSN-L)
- OpenAlex
- W2110405746
- Cited by
- 981
- References
- 22
- Citations by year
- 2026: 22 · 2025: 44 · 2024: 66 · 2023: 61 · 2022: 82 · 2021: 115 · 2020: 110 · 2019: 125 · 2018: 99 · 2017: 74 · 2016: 52 · 2015: 61 · 2014: 38 · 2013: 25 · 2012: 7
- FWCI
- 78.04
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- Proceedings - IEEE International Conference on Robotics and Automation/Proceedings (publishedVersion)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Robotics and Sensor-Based Localization
- Physical Sciences › Engineering › Automotive Engineering › Autonomous Vehicle Technology and Safety
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Image and Video Retrieval Techniques
- Keywords
- visual SLAM, visual navigation, route recognition, seasonal variation, heavy rainfall, vehicle-mounted camera, trajectory matching
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.