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 501–600 of 987
Visual localization under appearance change: filtering approaches
Doan, A.-D., Latif, Y., Chin, T.-J., Liu, Y., Chng, S.-F., Do, T.-T., & Reid, I. D. (2020) · Neural Computing and Applications · doi:10.1007/s00521-020-05339-y · Cited by 10
“An orthogonal approach to address appearance change, pioneered by SeqSLAM [26], is to consider sequence-to-sequence matching instead of matching a single query image to previously observed images.”
BackgroundLandmark Generation in Visual Place Recognition Using Multi-Scale Sliding Window for Robotics
Yang, B., Xu, X., Li, J., & Zhang, H. (2019) · Applied Sciences · doi:10.3390/app9153146 · Cited by 11
“Due to enjoying better discrimination capacity of the ConvNet feature, massive efforts are devoted to using it for replacing the traditional hand-crafted global features (e.g., SAD features in SeqSLAM [25] and HOG features in [26]) and thus achieve higher recognition accuracies.”
MethodNowicki, M., & Skrzypczyński, P. (2019) · Sensors · doi:10.3390/s19173657 · Cited by 9
“Another possibility is to process sequences of images, as proposed in SeqSLAM [33].”
MethodTowards combining a neocortex model with entorhinal grid cells for mobile robot localization
Schubert, S., Neubert, P., & Protzel, P. (2019) · European Conference on Mobile Robots · doi:10.1109/ecmr.2019.8870939 · Cited by 10
“SMART [17] extends the sequential image processing approach from SeqSLAM [12] with additional odometry input.”
MethodWeighted Node Mapping and Localisation on a Pixel Processor Array
Castillo-Elizalde, H., Liu, Y., Bose, L., & Mayol-Cuevas, W. W. (2021) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48506.2021.9561524 · Cited by 9
“SeqSLAM [12] is a topological version of a SLAM algorithm that uses the idea of sub-routes and low resolution images.”
BackgroundMethodIncreasing the efficiency of 6-DoF visual localization using multi-modal sensory data
Clark, R., Wang, S., Wen, H., Trigoni, N., & Markham, A. (2016) · IEEE-RAS International Conference on Humanoid Robots (Humanoids) · doi:10.1109/humanoids.2016.7803390 · Cited by 9
Survey of 3D Map in SLAM: Localization and Navigation
Yang, A., Luo, Y., Chen, L., & Xu, Y. (2017) · Communications in computer and information science · doi:10.1007/978-981-10-6370-1_41 · Cited by 10
Semantically-Aware Attentive Neural Embeddings for Image-based Visual LocalizationPreprint
Seymour, Z., Sikka, K., Chiu, H.-P., Samarasekera, S., & Kumar, R. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1812.03402 · Cited by 11
“Most initial works for visual localization relied on matching Bag of Visual Word (BoVW)-type features [5, 20] or using global image descriptors in addition to sequential search [33].”
MethodSuperquadric Object Representation for Optimization-based Semantic SLAMPreprint
Tschopp, F., Nieto, J. I., Siegwart, R., & Cadena, C. (2021) · arXiv (Cornell University) · doi:10.3929/ethz-b-000487527 · Cited by 12
“Such changes are especially evident in outdoor applications, in which daytime, weather, and seasonal conditions can influence the appearance significantly [10], [11].”
BackgroundTan, Y., Ji, P., Zhang, Y., Ge, F., & Zhu, S. (2024) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2024.109425 · Cited by 8
CaseVPR: Correlation-Aware Sequential Embedding for Sequence-to-Frame Visual Place Recognition
Li, H., Peng, G., Zhang, J., Wen, M., Ma, Y., & Wang, D. (2025) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2025.3541452 · Cited by 8
Improved Seq SLAM for Real-Time Place Recognition and Navigation Error Correction
Wang, Y., Hu, X., Lian, J., Zhang, L., & Kong, X. (2015) · International Conference on Intelligent Human-Machine Systems and Cybernetics · doi:10.1109/ihmsc.2015.23 · Cited by 8
“Their approach, coined SeqSLAM, seeks to match sequence of images instead of matching single image[18].”
BackgroundMethodCombining Multiple Image Descriptions for Loop Closure Detection
Wang, X., Peng, G., & Zhang, H. (2017) · Journal of Intelligent & Robotic Systems · doi:10.1007/s10846-017-0755-7 · Cited by 7
“SeqSLAM [8] measures the similarity between image sequences instead of individual images.”
BackgroundCAMAL: Context-Aware Multi-scale Attention framework for Lightweight Visual Place Recognition
Khaliq, A., Ehsan, S., Milford, M. J., & McDonald-Maier, K. D. (2019) · Open Access at Essex (University of Essex) · Cited by 11
Self-citationConvolutional Neural Networks for Geo-Localisation with a Single Aerial Image
Cabrera-Ponce, A. A., & Martínez-Carranza, J. (2022) · Journal of Real-Time Image Processing · doi:10.1007/s11554-022-01207-1 · Cited by 8
“In a similar fashion, other works have used sequential descriptors to recognise and calculate the position by regression and matching [33].”
Teach and Repeat Navigation: A Robust Control Approach
Nourizadeh, P., Milford, M. J., & Fischer, T. (2024) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra57147.2024.10611662 · Cited by 8
Self-citationImage-based global localization using VG-RAM Weightless Neural Networks
Junior, L. J. L., Oliveira-Santos, T., Forechi, A., Veronese, L. D. P., Badue, C. S., & Souza, A. F. D. (2014) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn.2014.6889888 · Cited by 10
“SeqSLAM [15] is another state-of-the-art appearance-based SLAM that calculates the best candidate matching of an image within a segment of a sequence of previously seen images.”
BackgroundPlace Recognition in Gardens by Learning Visual Representations: Data Set and Benchmark Analysis
Leyva-Vallina, M., Strisciuglio, N., & Petkov, N. (2019) · Lecture notes in computer science · doi:10.1007/978-3-030-29888-3_26 · Cited by 11
Communication constrained cloud-based long-term visual localization in real time
Ding, X., Wang, Y., Tang, L., Yin, H., & Xiong, R. (2019) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros40897.2019.8968550 · Cited by 8
“In order to improve the robustness of localization, there is another line of methods that aggregates the local observations to generate larger field of view for localization [8, 9].”
BackgroundMethodA Robust Image-Sequence-Based Framework for Visual Place Recognition in Changing Environments
Wang, Y., Xue, T., & Li, Q. (2020) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2020.2977128 · Cited by 8
Li, C.-H. G., Hong, Y.-F., Hsu, P.-K., & Maneewarn, T. (2020) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2020.103578 · Cited by 8
A Panoramic Localizer Based on Coarse-to-Fine Descriptors for Navigation Assistance
Fang, Y., Yang, K., Cheng, R., Sun, L., & Wang, K. (2020) · Sensors · doi:10.3390/s20154177 · Cited by 8
“Global descriptors are usually extracted from the whole image to represent the holistic scene [20,28].”
BackgroundBeyond ANN: Exploiting Structural Knowledge for Efficient Place Recognition
Schubert, S., Neubert, P., & Protzel, P. (2021) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48506.2021.9561006 · Cited by 10
“, [8][9]) perform a full pairwise comparison between all database and query descriptors.”
MethodJaenal, A., Moreno, F.-Á., & González-Jiménez, J. (2021) · Sensors · doi:10.3390/s21072483 · Cited by 8
“Building upon SeqSLAM, SMART [11] integrates odometry readings to provide more consistent results.”
MethodLFM: A Lightweight LCD Algorithm Based on Feature Matching between Similar Key Frames
Zhu, Z., Xu, X., Liu, X., & Jiang, Y. (2021) · Sensors · doi:10.3390/s21134499 · Cited by 8
“proposed a sequence-based approach that calculates the best candidate matching positions in each local navigation sequence and then achieves localization by identifying the coherent sequences of these matches [8].”
BackgroundTakeda, K., & Kanji, T. (2021) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48506.2021.9561158 · Cited by 8
Hierarchical Visual Place Recognition Based on Semantic-Aggregation
Chen, B., Song, X., Shen, H., & Lü, T. (2021) · Applied Sciences · doi:10.3390/app11209540 · Cited by 10
Soft Manipulator Fault Detection and Identification Using ANC-based LSTM
Gu, H., Hu, H., Wang, H., & Chen, W. (2021) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros51168.2021.9636868 · Cited by 8
“To reduce the disturbance caused by false positive prediction, we adopt a sequence-based pipeline like other sequential methods [17].”
MethodDrive on Pedestrian Walk. TUK Campus Dataset
Keen, H. E., Jan, Q. H., & Berns, K. (2021) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros51168.2021.9636848 · Cited by 8
“Authors in [7] provide the vision data for place recognition in the day as well as in the night.”
BackgroundBenchmarking Visual Localization for Autonomous Navigation
Suomela, L., Kalliola, J., Dag, A., Edelman, H., & Kämäräinen, J. (2023) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv56688.2023.00296 · Cited by 11
“We limit our experiments to the state-of-the-art hierarchical localization methods, and do not consider other approaches such as direct image alignment [61, 53] or sequence-based methods [39, 42].”
MethodLearning Deep Representation for Place Recognition in SLAM
Mukherjee, A., Chakraborty, S., & Saha, S. K. (2017) · Lecture notes in computer science · doi:10.1007/978-3-319-69900-4_71 · Cited by 8
Application of a Bio-inspired Localization Model to Autonomous Vehicles
Espada, Y., Cuperlier, N., Bresson, G., & Romain, O. (2018) · International Conference on Control, Automation, Robotics and Vision (ICARCV) · doi:10.1109/icarcv.2018.8581268 · Cited by 8
“0 [9], SeqSLAM [10] or SLAM across seasons [11], exhibit a good ability to handle dynamic environments but can be difficult to apply online or do not scale well due to the necessity to compare the current image with a whole database.”
BackgroundInfluentialBibars, A., & Mahroos, M. (2019) · IET Computer Vision · doi:10.1049/iet-cvi.2018.5203 · Cited by 8
“SeqSLAM [2] is among the robust visual place recognition approaches under appearance changes which use whole image matching.”
BackgroundMethodResultInfluentialPlace Recognition Using Line-Junction-Lines in Urban Environments
Tang, X., Fu, W., Jiang, M., Peng, G., Wu, Z., Yue, Y., & Wang, D. (2019) · IEEE Conference on Cybernetics and Intelligent Systems · doi:10.1109/cis-ram47153.2019.9095776 · Cited by 8
“To overcome the low tolerance to illumination and viewpoint change, SeqSLAM [9] proposed to recognize consecutive images instead of searching for single frame.”
BackgroundLocNet: Global localization in 3D point clouds for mobile robots.Preprint
Yin, H., Wang, Y., Tang, L., Ding, X., & Xiong, R. (2017) · arXiv (Cornell University) · Cited by 8
Visual SLAM with Network Uncertainty Informed Feature Selection.Preprint
Ganti, P., & Waslander, S. L. (2018) · arXiv (Cornell University) · Cited by 8
Visual Place Recognition for Simultaneous Localization and Mapping
Tsintotas, K. A., Bampis, L., & Gasteratos, A. C. (2022) · Wiley eBooks · doi:10.1002/9781394152636.ch4 · Cited by 7
Hyperion – A Fast, Versatile Symbolic Gaussian Belief Propagation Framework for Continuous-Time SLAM
Hug, D., Alzugaray, I., & Chli, M. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-73404-5_13 · Cited by 14
Eyes in the Back of Your Head: Robust Visual Teach & Repeat Using Multiple Stereo Cameras
Paton, M., Pomerleau, F., & Barfoot, T. D. (2015) · Canadian Conference on Computer and Robot Vision · doi:10.1109/crv.2015.16 · Cited by 8
“Recent methods that address appearance change are not suitable for this requirement, providing only topological localization [3] or requiring offline collection of the scene in multiple appearances [4, 5].”
MethodPHROG: A Multimodal Feature for Place Recognition
Bonardi, F., Aïnouz, S., Boutteau, R., Dupuis, Y., Savatier, X., & Vasseur, P. (2017) · Sensors · doi:10.3390/s17051167 · Cited by 7
“Some other authors choose as well global description of the images, called image signatures, and compare directly these compact representations [6,8].”
BackgroundDeep Embeddings-based Place Recognition Robust to Motion Blur
Woźniak, P., & Kwolek, B. (2021) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw54120.2021.00203 · Cited by 7
“Instead of using individual images or image descriptors a correlation-based matching on entire sequences was introduced in [22].”
Towards bio-inspired place recognition over multiple spatial scales
Chen, Z., Jacobson, A., Erdem, U. M., Hasselmo, M. E., & Milford, M. J. (2013) · QUT ePrints (Queensland University of Technology) · Cited by 7
Self-citationCarson, H., Ford, J., & Milford, M. J. (2023) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48891.2023.10160679 · Cited by 7
“Sequential filtering can also be used to exploit the rich spatio-temporal information provided as a localization system moves along a route, rather than treating each image independently [6], including learned descriptors based on contrastive learning methods such as SeqNet [19] and SeqMatchNet [20].”
BackgroundMethodInfluentialSelf-citationZhou, B., Xiao, Y., Li, Q., Sun, C., Wang, B., Pan, L., Zhang, D., Zhu, J., & Li, Q. (2024) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2024.3350043 · Cited by 10
“SeqSlam exploits the image sequence to achieve place recognition under challenging weather and illumination conditions [7].”
BackgroundXu, H., Liu, H., Meng, S., & Sun, Y. (2023) · IEEE International Conference on Intelligent Transportation Systems (ITSC) · doi:10.1109/itsc57777.2023.10421887 · Cited by 7
“Sequence matching is a general paradigm for sequence-based place recognition [17], [18], which compares each frame of the input sequence with all images in the database, and finds the place corresponding to the sequence through similarity score aggregation.”
BackgroundAbsolute geo-localization thanks to Hidden Markov Model and exemplar-based metric learning
Barz, C. L., Thome, N., Cord, M., Herbin, S., & Sanfourche, M. (2015) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw.2015.7301394 · Cited by 7
“…acquired at different times and with different cameras may show huge appearance differences due to illumination and colorimetry variations (e.g. sunny 1 or cloudy days), camera viewpoints changes, scene modifications (e.g. seasonal changes, building construction) and occlusions (e.g. by cars) (Fig.”
BackgroundPerformance evaluation of whole-image descriptors in visual loop closure detection
Liu, Y., & Zhang, H. (2013) · IEEE International Conference on Information and Automation (ICIA) · doi:10.1109/icinfa.2013.6720388 · Cited by 7
“In SeqSLAM [18], the image descriptor used is the simplest one, a downsampled image, with the results that can be surprisingly good if image sequences are matched.”
MethodSceneSLAM: A SLAM framework combined with scene detection
Tong, Z., Shi, D., & Yang, S. (2017) · IEEE International Conference on Robotics and Biomimetics (ROBIO) · doi:10.1109/robio.2017.8324464 · Cited by 7
“matching appearance-based sequences [29], [30] and data-based motion removal approach [31]”
Robust ConvNet Landmark-Based Visual Place Recognition by Optimizing Landmark Matching
Kong, Y., Liu, W., & Chen, Z. (2019) · IEEE Access · doi:10.1109/access.2019.2901984 · Cited by 7
“Sequence search-based idea was first introduced to VPR by SeqSLAM in [27].”
From Neurorobotic Localization to Autonomous Vehicles
Espada, Y., Cuperlier, N., Bresson, G., & Romain, O. (2019) · Unmanned Systems · doi:10.1142/s2301385019410048 · Cited by 7
Keyframe-Based Camera Relocalization Method Using Landmark and Keypoint Matching
Yang, B., Xu, X., & Li, J. (2019) · IEEE Access · doi:10.1109/access.2019.2925121 · Cited by 7
“[27] M. J. Milford and G. F. Wyeth, ‘‘SeqSLAM: Visual route-based navigation for sunny summer days and stormy winter nights,’’ in Proc.”
MethodRobust Outdoor Self-localization In Changing Environments
Haris, M., Franzius, M., & Bauer-Wersing, U. (2019) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros40897.2019.8967549 · Cited by 7
“The use of image sequence matching has shown significant improvement in visual localization even with drastic appearance changes [21].”
BackgroundVisual Sequence Place Recognition with Improved Dynamic Time Warping
Lu, F. H., Chen, B., Guo, Z., & Zhou, X. (2019) · IEEE International Conference on Tools with Artificial Intelligence (ICTAI) · doi:10.1109/ictai.2019.00145 · Cited by 7
“We compared the midpoint of the recognized sequence with the midpoint of the current sequence, which is the same as SeqSLAM study.”
MethodInfluentialCityLearn: Diverse Real-World Environments for Sample-Efficient Navigation Policy Learning
Chancán, M., & Milford, M. J. (2020) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra40945.2020.9197336 · Cited by 8
“namic objects, such as cars, traffic, and pedestrians, along with longer-term changes such as construction or roadworks [37], [44], [60]–[67] (Fig.”
BackgroundMethodInfluentialSelf-citationTwo-Stage vSLAM Loop Closure Detection Based on Sequence Node Matching and Semi-Semantic Autoencoder
Wang, Z., Peng, Z., Guan, Y. L., & Wu, L. (2021) · Journal of Intelligent & Robotic Systems · doi:10.1007/s10846-020-01302-0 · Cited by 7
“A pure topological map contains information about the relative location of the landmarks, but does not store metric information on the degree of relevance of these locations [23].”
BackgroundResolving Place Recognition Inconsistencies Using Intra-Set Similarities
Neubert, P., Schubert, S., & Protzel, P. (2021) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2021.3060729 · Cited by 8
“The back-end of SeqSLAM [2] is a post-processing of the pairwise similarity matrix between database and query sets in order to find short linear sequences of neighbored high similarities.”
BackgroundMethodInfluentialA novel fusing semantic- and appearance-based descriptors for visual loop closure detection
Wu, P., Wang, J., Wang, C., Zhang, L., & Wang, Y. (2021) · Optik · doi:10.1016/j.ijleo.2021.167230 · Cited by 7
Sequential appearance-based Global Localization using an ensemble of kNN-DTW classifiers
Forechi, A., Souza, A. F. D., Badue, C. S., & Oliveira-Santos, T. (2016) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn.2016.7727550 · Cited by 7
“SeqSLAM is yet another appearance-based SLAM that calculates the best candidate matching of an image within a segment of a sequence of previously seen images.”
MethodImage-based localization using Gaussian processes
López-Antequera, M., Petkov, N., & González-Jiménez, J. (2016) · International Conference on Indoor Positioning and Indoor Navigation · doi:10.1109/ipin.2016.7743697 · Cited by 7
Compact Environment-Invariant Codes for Robust Visual Place Recognition
Jain, U., Namboodiri, V. P., & Pandey, G. (2017) · Canadian Conference on Computer and Robot Vision · doi:10.1109/crv.2017.22 · Cited by 7
“VPR is similar to image retrieval and research in VPR [4], [8], [10], [11], [13] plot PR curves by altering a parameter.”
BackgroundMethodInfluentialA generalized multi-snapshot model for 3D homing and route following
Differt, D., & Stürzl, W. (2020) · Adaptive Behavior · doi:10.1177/1059712320911217 · Cited by 7
“While the limited search range already reduces computation time and the chance of mis-matches, further improvements might be possible by enforcing a stricter sequence-based matching as in the literature (e.g. Grixa et al., 2018; Milford & Wyeth, 2012; Pepperell et al., 2014).”
MethodGSAP: A Global Structure Attention Pooling Method for Graph-Based Visual Place Recognition
Yang, Y., Ma, B., Liu, X., Zhao, L., & Huang, S. (2021) · Remote Sensing · doi:10.3390/rs13081467 · Cited by 7
“In the VPR literature, various feature extraction methods have been developed for visual place recognition, including deep convolutional feature-based methods [6–13], handicraft feature-based methods [2,18], semantic information-based methods [19–25], sequence-based methods [26,27], and graph-based methods [19,20,28–3…”
MethodA Flexible and Efficient Loop Closure Detection Based on Motion Knowledge
Liu, B., Tang, F., Fu, Y., Yang, Y., & Wu, Y. (2021) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48506.2021.9561126 · Cited by 7
Feature Map Filtering: Improving Visual Place Recognition with Convolutional CalibrationPreprint
Hausler, S., Jacobson, A., & Milford, M. J. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1810.12465 · Cited by 7
“We calculate the quality of the best matching template using a method originally proposed in SeqSLAM [Milford and Wyeth, 2012], where the quality score is the ratio between the score at the best matching template and the next highest score outside a window around the best matching template.”
MethodSelf-citationSelf-Supervised Visual Place Recognition Learning in Mobile RobotsPreprint
Pillai, S., & Leonard, J. J. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1905.04453 · Cited by 7
“In SeqSLAM, the authors [29] identify potential loop closures by casting it as a sequence alignment problem, while others [9] rely on temporal consistency checks across long image sequences in order to robustly propose loop closures.”
BackgroundArshad, S., & Park, T.-H. (2024) · Sensors · doi:10.3390/s24030906 · Cited by 6
“Ground truth information is established based on GPS data, with a match considered a true positive if it falls within a 10 m range, following a thresholding approach similar to [49,50].”
MethodAdaptive Route Memory Sequences for Insect-Inspired Visual Route Navigation
Kagioulis, E., Knight, J. C., Graham, P., Nowotny, T., & Philippides, A. O. (2024) · Biomimetics · doi:10.3390/biomimetics9120731 · Cited by 6
iLoc: An Adaptive, Efficient, and Robust Visual Localization System
Yin, P., Zhao, S., Wang, J., Ge, R., Ji, J., Hu, Y., Liu, H., & Han, J. (2025) · IEEE Transactions on Robotics · doi:10.1109/tro.2025.3530273 · Cited by 7
“However, evaluating P ( z r 1: T | x T , M ) is time consuming when the agent tries to locate over large-scale scenery, similar to the original burst-force searching manner of global matching [34].”
InfluentialUnpaired image-to-image translation with content preserving perspective: a review
Safayani, M., Mirzapour, B., Aghaebrahimiyan, H., Salehi, N., & Ravaee, H. (2026) · Multimedia Tools and Applications · doi:10.1007/s11042-026-21698-5 · Cited by 6
Towards condition-invariant, top-down visual place recognition
Milford, M. J., Vig, E., Scheirer, W. J., & Cox, D. (2013) · QUT ePrints (Queensland University of Technology) · Cited by 6
Self-citationLPD-AE: Latent Space Representation of Large-Scale 3D Point Cloud
Suo, C., Liu, Z., Mo, L., & Liu, Y. (2020) · IEEE Access · doi:10.1109/access.2020.2999727 · Cited by 6
“16, and the blue keyframes represent the corresponding loop closure paths detected, like [72].”
BackgroundLoop Closure Detection Based on Image Semantic Segmentation in Indoor Environment
Li, J., Wang, P., Ni, C., & Rong, W. (2022) · Mathematical Problems in Engineering · doi:10.1155/2022/7765479 · Cited by 6
“…the traditional loop closure detection method generally abstracts image information and extracts the artificially designed image features [6–8] as image descriptors to represent the image. commonly used algorithms can be divided into two categories: bag-of-words (BoW) [9] and global…”
MethodUnsupervised Appearance Map Abstraction for Indoor Visual Place Recognition With Mobile Robots
Jaenal, A., Moreno, F.-Á., & González-Jiménez, J. (2022) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2022.3186768 · Cited by 6
“We compare the mean localization performance on both COLD evaluation sequences employing NetVLAD, using our pVPR localization on a map with M = 35 against IR and two online available state-of-the-art methods: the topological filter from [24] with NetVLAD descriptors and SeqSLAM[7] with images subsampled to 48×64.”
BackgroundMethodInfluentialLooking Beneath More: A Sequence-based Localizing Ground Penetrating Radar Framework
Zhang, P., Zhi, S., Yuan, Y., Bi, B., Xin, Q., Huang, X., & Shen, L. C. (2024) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra57147.2024.10610174 · Cited by 6
“2) Leveraging sequential information to enhance performance, as demonstrated by methods like SeqSLAM, Seq-MatchNet, SeqVlad, and others [19]–[22], which take the continuity of robot motion into account.”
BackgroundMethodRadarLCD: Learnable Radar-based Loop Closure Detection Pipeline
Usuelli, M., Frosi, M., Cudrano, P., Mentasti, S., & Matteucci, M. (2024) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn60899.2024.10651220 · Cited by 7
“Gadd et al. [17] improved upon this approach by incorporating sequence-based place recognition using SeqSLAM [39].”
MethodZhou, J., Shen, J., Lyu, S., & Lu, Y. (2024) · Expert Systems with Applications · doi:10.1016/j.eswa.2024.125659 · Cited by 6
DK-SLAM: Monocular Visual SLAM with Deep Keypoint Learning, Tracking, and Loop Closing
Qu, H., Zhang, L., Mao, J., Tie, J., He, X., Hu, X., Shi, Y., & Chen, C. (2025) · Applied Sciences · doi:10.3390/app15147838 · Cited by 6
Toward mutual information based place recognition
Pandey, G., McBride, J. R., Savarese, S., & Eustice, R. M. (2014) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2014.6907317 · Cited by 6
“[11] proposed to match sequences of images instead of a single image and showed good precision in recognizing places across different seasons (summer-rain).”
BackgroundLoop Closure Detection by Algorithmic Information Theory: Implemented on Range and Camera Image Data
Ravari, A. N., & Taghirad, H. D. (2014) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2014.2300180 · Cited by 6
An assistive haptic interface for appearance-based indoor navigation
Rivera-Rubio, J., Arulkumaran, K., Hemang, R., Alexiou, I., & Bharath, A. A. (2016) · Computer Vision and Image Understanding · doi:10.1016/j.cviu.2016.02.014 · Cited by 6
An adaptive localization system for image storage and localization latency requirements
Mao, J., Hu, X., & Milford, M. J. (2018) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2018.06.007 · Cited by 6
“Image comparison parameters were similar to 12 OpenSeqSLAM [5].”
BackgroundMethodResultInfluentialSelf-citationHM$^4$: Hidden Markov Model With Memory Management for Visual Place Recognition
Doan, A.-D., Latif, Y., Chin, T.-J., & Reid, I. D. (2020) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2020.3036615 · Cited by 8
“To address this issue, temporal continuity [13] and appearance transfer [14] have been shown to be useful, but those did not consider lifelong operation where appearance changes occur continuously.”
BackgroundMount, J., Xu, M., Dawes, L., & Milford, M. (2021) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2020.3043171 · Cited by 6
“The results additionally demonstrate that such a system could be used in VPR systems like SeqSLAM [2] to optimise the image resolution for the deployment environment.”
MethodSelf-citationSelf-Supervised Map-Segmentation by Mining Minimal-Map-Segments
Kanji, T. (2020) · IEEE Intelligent Vehicles Symposium · doi:10.1109/iv47402.2020.9304724 · Cited by 6
“The developed VPC systems were evaluated in challenging cross-season VPR scenarios [44], using the publicly available NCLT dataset [45].”
BackgroundMethodTomită, M.-A., Zaffar, M., Ferrarini, B., Milford, M. J., McDonald-Maier, K. D., & Ehsan, S. (2022) · IEEE Access · doi:10.1109/access.2022.3196389 · Cited by 6
“that sequential-filtering mostly helps with introducing conditional-invariance [1], however, the results obtained on Gardens Point day-to-day shown in Fig.”
BackgroundResultInfluentialSelf-citationVariational Bayesian Approach to Condition-Invariant Feature Extraction for Visual Place Recognition
Oh, J., & Eoh, G. (2021) · Applied Sciences · doi:10.3390/app11198976 · Cited by 6
“Existing handcraft features such as SAD and FAB–MAP showed they are not suitable for place recognition in a changing environment.”
MethodA Lightweight sequence-based Unsupervised Loop Closure Detection
Xiong, F., Ding, Y., Yu, M., Zhao, W., Zheng, N., & Ren, P. (2021) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn52387.2021.9534180 · Cited by 6
“Meanwhile, using multiple images in sequence for similarity measurement can overcome the influence of noise and provide more stable and accurate results [15]–[17].”
BackgroundMethodStable keypoints selection for 2D LiDAR based place recognition with map data reduction
Douadi, L., Dupuis, Y., & Vasseur, P. (2022) · Robotica · doi:10.1017/s0263574722000613 · Cited by 6
“[9], models the appearance with image sequences.”
Fast Scan Context Matching for Omnidirectional 3D Scan
Kihara, H., Kumon, M., Nakatsuma, K., & Furukawa, T. (2022) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros47612.2022.9981143 · Cited by 6
“For example, FAB-MAP incorporated the bag-of-words approach to store the appearance keys[2], a sequence of images was used to match in SeqSLAM[3] and BRIEF feature[4] was accepted as a descriptor in BRIEF-Gist[5].”
MethodTrajectory Tracking via Multiscale Continuous Attractor Networks
Joseph, T., Fischer, T., & Milford, M. J. (2023) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros55552.2023.10341938 · Cited by 6
“While we used ATE to evaluate the performance of our networks, we utilized the Sum of Absolute Differences (SAD) method [41] during the tuning process to measure the fitness of the networks.”
MethodSelf-citationVG-SSL: Benchmarking Self-Supervised Representation Learning Approaches for Visual Geo-Localization
Xiao, J., Zhu, G., & Loianno, G. (2025) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv61041.2025.00649 · Cited by 9
“Traditional VG approaches [17, 18, 24, 33] rely on handcrafted templates, which are generally less resilient to noise and appearance changes (e.g., illumination).”
Hafez, A. H. A., Arora, M., Krishna, K. M., & Jawahar, C. V. (2015) · Advanced Robotics · doi:10.1080/01691864.2015.1090336 · Cited by 6
Sequence-based multimodal apprenticeship learning for robot perception and decision making
Han, F., Yang, X., Zhang, Y., & Zhang, H. (2017) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2017.7989301 · Cited by 6
DashCam video compression using historical data
Ma, B., & Reibman, A. R. (2016) · Picture Coding Symposium · doi:10.1109/pcs.2016.7906326 · Cited by 6
“relative to the matching algorithm in [14].”
BackgroundMethodInfluentialVisual place recognition with CNNs: From global to partial
Xin, Z., Cui, X., Zhang, J., Yang, Y., & Wang, Y. (2017) · International Conference on Image Processing Theory, Tools and Applications (IPTA) · doi:10.1109/ipta.2017.8310121 · Cited by 6
“Instead of matching based on single images, SeqSLAM [19] improved the precision by using sequence search techniques.”
MethodLeón, J., Farfan, J., & Arapa, E. G. C. (2018) · IEEE International Conference on Electronics, Electrical Engineering and Computing (INTERCON) · doi:10.1109/intercon.2018.8526392 · Cited by 6
“Also, to further improve the robustness in front of appearance change, they developed a descriptor normalization scheme that builds on the success of normalization schemes for pure appearance-based techniques such as SeqSLAM [8].”
MethodEstimation and navigation methods with limited information for autonomous urban driving
Chipka, J. B., & Campbell, M. (2019) · Engineering Reports · doi:10.1002/eng2.12054 · Cited by 7
“18 Although both methods do not rely on GPS measurements or an a priori HD map, they operate purely in appearance space and make no attempt to localize the vehicle in metric coordinates, either global or map based; in other words, these methods act as scene-matching algorithms.”
MethodBio-inspired multi-scale fusion
Hausler, S., Chen, Z., Hasselmo, M. E., & Milford, M. J. (2020) · Biological Cybernetics · doi:10.1007/s00422-020-00831-z · Cited by 8
“In this work, we use the well-established image processing methods of Sum of Absolute Differences with Patch Normalization (Milford and Wyeth 2012), Gist (Oliva and Torralba 2006), and NetVLAD (Arandjelovic et al. 2018), which collectively span the range of technique types used in the field of…”
MethodInfluentialSelf-citationImproved Visual Localization via Graph Filtering
Lassance, C., Latif, Y., Garg, R., Gripon, V., & Reid, I. D. (2021) · Journal of Imaging · doi:10.3390/jimaging7020020 · Cited by 7
Garcia‐Fidalgo, E., & Ortiz, A. (2022) · Autonomous Robots · doi:10.1007/s10514-021-10032-7 · Cited by 6
“These methods are globally referred to as appearance-based LCD (Angeli et al. 2008; Milford and Wyeth 2012; Cummins and Newman 2011; Galvez-López and Tardos 2012; Khan and Wollherr 2015; Garcia-Fidalgo and Ortiz 2018; Bampis et al. 2018).”
BackgroundMethod
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.