Object recognition from local scale-invariant features

By David G. Lowe. 1999

Subfields: 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 Computer Vision and Pattern Recognition

Citations in the corpus

Do Better ImageNet Models Transfer Better? (2018, neutral): “Before the rise of convolutional neural networks, most approaches to image understanding relied on hand-engineered feature descriptors [45, 15, 3].”
Making Convolutional Networks Shift-Invariant Again (2019, methodology): “Classic hand-engineered computer vision and image processing representations, such as SIFT (Lowe, 1999), wavelets, and image pyramids (Adelson et al., 1984; Burt & Adelson, 1987) also extract features in a sliding window manner, often with some subsampling factor.”
Challenges in representation learning: A report on three machine learning contests (2013, neutral): “Their approach used SIFT [15] and MKL.”
Colorful Image Colorization (2016, background): “Deshpande et al. [10] train a Gaussian random ﬁeld on SIFT features [11] on a small set of 240 images from 6 scene categories.”
You Only Look Once: Unified, Real-Time Object Detection (2015, methodology): “Detection pipelines generally start by extracting a set of robust features from input images (Haar [25], SIFT [23], HOG [4], convolutional features [6]).”
Self-Supervised Learning of Pretext-Invariant Representations (2019, background): “In fact, invariance is one of the core tenets of designing ‘good’ features [8, 45, 48].”
Describing Textures in the Wild (2013, methodology): “The ﬁrst one is the use of the SIFT descriptors, originally developed for object matching [20], that are more distinctive that local descriptors popular in texture analysis such as ﬁlter banks [13, 19, 36], local intensity patterns [23], and patches [35].”
Generating Videos with Scene Dynamics (2016, methodology): “ (babies), beaches, and train station. Stabilization: As we are interested in the movement of objects and not camera shake, we stabilize the camera motion for both datasets. We extract SIFT keypoints [21], use RANSAC to estimate a homography (rotation, translation, scale) between adjacent frames, an”
Representation Learning: A Review and New Perspectives (2012, methodology): “The most prevalent approach to incorporating prior knowledge is to hand-design better features to feed a generic classifier, and has been used extensively in computer vision (e.g. (Lowe, 1999)).”
Neural Architecture Search with Reinforcement Learning (2016, methodology): “, from SIFT (Lowe, 1999), and HOG (Dalal & Triggs, 2005), to AlexNet (Krizhevsky et al.”
Learning Generative Visual Models from Few Training Examples: An Incremental Bayesian Approach Tested on 101 Object Categories (2004, background): “Signiﬁcant progress has been made on the issues of representation of objects [17], [18] and object categories [2], [3], [5]–[8] with a broad agreement for models that are composed of ‘parts’ (textured patches, features) and ‘geometry’ (or mutual position of the parts).”
Deformable Convolutional Networks (2017, methodology): “Notable examples include scale invariant feature transform (SIFT) [42] and ORB [49] (O for orientation).”
Learning to See by Moving (2015, methodology): “Using this insight, we pose the problem of correlating visual stimuli with egomotion as the problem of predicting the camera transformation from the consequent pairs of images that the agent receives while it moves.”
Joint Training of a Convolutional Network and a Graphical Model for Human Pose Estimation (2014, background): “We show that joint training of these two model paradigms improves performance and allows us to significantly outperform existing state-of-the-art techniques.”
Understanding deep image representations by inverting them (2014, methodology): “These are equivalent to Lowe’s [19] SIFT and the DPM V5 HOG [7, 8].”
Unsupervised Learning by Predicting Noise (2017, methodology): “Using this pre-processing is not surprising since most hand-made features like SIFT or HoG are based on image gradients (Lowe, 1999; Dalal & Triggs, 2005).”
Why do deep convolutional networks generalize so poorly to small image transformations? (2018, methodology): “Alternatively, one could use specially designed features such as SIFT in which invariance is hard coded [7, 29, 30, 31] or neural network architectures that explicitly enforce invariance [36, 19, 5, 6, 11, 12, 47, 46, 8].”
Differentially Private Learning Needs Better Features (or Much More Data) (2020, methodology): “We focus on computer vision, a canonical domain for private deep learning (Abadi et al., 2016; Yu et al., 2019b; Papernot et al., 2020b; Nasr et al., 2020)), with a rich literature on handcrafted features (Lowe, 1999; Dalal & Triggs, 2005; Bruna & Mallat, 2013).”