Place Recognition with ConvNet Landmarks: Viewpoint-Robust, Condition-Robust, Training-Free

By Niko Sünderhauf, Sareh Abolahrari Shirazi, Adam Jacobson, Feras Dayoub, Edward Pepperell, Ben Upcroft, Michael Milford. 2015

409 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Aerospace Engineering

In Computer Vision and Pattern Recognition

Citations in the corpus

NetVLAD: CNN Architecture for Weakly Supervised Place Recognition (2015, background): “While there have been many improvements in designing better image retrieval [22], [23], [24], [25], [48], [49], [50], [51], [52], [53], [54], [55], [56], [57], [58], [59], [60], [61], [62] and place recognition [2], [3], [4], [5], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16] systems,”
Benchmarking 6DOF Outdoor Visual Localization in Changing Conditions (2018, methodology): “Learning-based localization has been proposed to solve both loop-closure detection [17, 45, 64, 66] and pose estimation [19, 31, 74].”
Street-view change detection with deconvolutional networks (2016, methodology): “image classification [20], semantic segmentation [21, 28] and place recognition [37]), we adopt a deep deconvolutional architecture [31] for the task of structural change detection.”
Risky Action Recognition in Lane Change Video Clips using Deep Spatiotemporal Networks with Segmentation Mask Transfer (2019, methodology): “It has been used for natural language processing [26], speech recognition across different languages [27], voice conversion [28] and other computer vision applications [29].”
Fusion and binarization of CNN features for robust topological localization across seasons (2016, methodology): “In works such as [25], only the features generated by the third convolutional layer (dconv3 ) are considered in the place description process, which produces a loss of invariance.”
Deep-learning in Mobile Robotics - from Perception to Control Systems: A Survey on Why and Why not (2016, background): “Application in place recognition for mobile robots were naturally motivated (Chen et al., 2014; S¨underhauf et al., 2015a).”
Self-Supervised Visual Descriptor Learning for Dense Correspondence (2017, methodology): “[23] use AlexNet features pre-trained on ImageNet for visual place recognition.”
Made to measure: Bespoke landmarks for 24-hour, all-weather localisation with a camera (2016, methodology): “used Edge Boxes to generate landmarks from each image [22][23].”
High-fidelity simulation for evaluating robotic vision performance (2016, background): “There are some place recognition methods which have proven to work well with both condition and viewpoint variations as in [11], [13] and [17].”
One-Shot Reinforcement Learning for Robot Navigation with Interactive Replay (2017, methodology): “Using a visual encoder network that has been trained on a large vision dataset can help compensate for the lack of variety in the training environment [21], so we use a ResNet-50 network trained on ImageNet [22, 23] as in [2] and extract the pre-final fully-connected layer as a 2048-dimensional rich”
On the performance of ConvNet features for place recognition (2015, methodology): “In subsequent work [33] we addressed the scenario of viewpoint and appearance changes occurring simultaneously using a number of new dataset.”
24/7 Place Recognition by View Synthesis (2018, positive): “Convolutional neural network (CNN) descriptors have been recently used for the robot re-localization task in [53] who extract object”