Consensus: A Paradigm for Model Fitting with Apphcatlons to Image Analysis and Automated Cartography

Subfields: Computer Vision and Pattern Recognition

Citations in the corpus

• Delving into Egocentric Actions (2015, neutral): “Our results demonstrate a significant performance boost on three existing datasets.”
• Stronger data poisoning attacks break data sanitization defenses (2018, methodology): “A different strategy is to learn multiple models on different (random or otherwise) subsets of data, in the hopes that the data will be relatively clean at least on some subsets (Fischler & Bolles, 1981; Kearns & Li, 1993; Cretu et al., 2008).”
• Sever: A Robust Meta-Algorithm for Stochastic Optimization (2018, methodology): “Popular methods include RANSAC [FB81], minimum covariance determinant [RD99], removal based on k-nearest neighbors [BKNS00], and Huberizing the loss [Owe07] (see [HA04] for a comprehensive survey).”
• Manipulating Machine Learning: Poisoning Attacks and Countermeasures for Regression Learning (2018, methodology): “TRIM performs significantly better and is much more effective in providing robustness compared to known methods from robust statistics (Huber [27] and RANSAC [18]), typically designed to provide resilience against noise and outliers.”
• Synthetic Data for Text Localisation in Natural Images (2016, methodology): “The normals are estimated automatically by first predicting a dense depth map using the CNN of [30] for the regions segmented above, and then fitting a planar facet to it using RANSAC [10].”
• Action recognition with trajectory-pooled deep-convolutional descriptors (2015, neutral): “RANSAC [8] algorithm to estimate homography matrix.”
• Understanding Image Representations by Measuring Their Equivariance and Equivalence (2014, neutral): “RANSAC (Fischler and Bolles 1981) in local feature pipelines or non-maxima-suppression for object detection tasks (Dalal and Triggs 2005; Girshick et al.”
• P-CNN: Pose-Based CNN Features for Action Recognition (2015, methodology): “A recent approach [37] removes trajectories consistent with the camera motion (estimated computing a homography using optical flow and SURF [1] point matches and RANSAC [14]).”
• Unsupervised Deep Homography: A Fast and Robust Homography Estimation Model (2017, methodology): “These feature pairs are then used to calculate the homography using RANSAC with a threshold of 5 pixels.”
• Fast image-based localization using direct 2D-to-3D matching (2011, methodology): “Besides the standard 6-point DLT algorithm for pose estimation (p6p), we also experiment with two other pose estimation strategies: (1) When an image’s focal length is available from the EXIF tag, we use the 3-point-pose solver by [7] to estimate the camera pose instead of the 6-point solver (p3p/p6”
• A Frequency Domain Approach to Registration of Aliased Images with Application to Super-resolution (2006, methodology): “They can be based on the whole image or on a set of selected corresponding feature vectors, as discussed by Capel and Zisserman [4] and by Fischler and Bolles in their RANSAC algorithm [23].”
• Evaluation of image-based modeling and laser scanning accuracy for emerging automated performance monitoring techniques (2011, background): “Due to the sensitivity of the reconstruction algorithm to these false matches, a fundamental matrix [21] within a RANSAC (random sample consensus) [34] iterative method is fit to each matching image pair.”
• Autoreject: Automated artifact rejection for MEG and EEG data (2016, methodology): “Finally, Figure 6D shows comparison to RANSAC.”
• Active Matching (2008, methodology): “When probabilistic priors are available, the unsatisfying random sampling and preset thresholds of RANSAC have been improved on by probabilistic methods such as the Joint Compatibility Branch and Bound (JCBB) algorithm [11] which matches features via a deterministic interpretation tree [7] and has b”
• Learning to Detect Objects with a 1 Megapixel Event Camera (2020, methodology): “Once feature points are obtained, we can fit an homography using standard methods, such as RANSAC [73].”