Unsupervised Feature Learning for RGB-D Based Object Recognition

By Liefeng Bo, Xiaofeng Ren, Dieter Fox. 2012

394 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition, Representation & Self-Supervised Learning

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

In Representation & Self-Supervised Learning

Citations in the corpus

Discriminative Unsupervised Feature Learning with Convolutional Neural Networks (2014, neutral): “To each feature map, we applied the pooling method that is commonly used for the respective dataset: 1) 4-quadrant max-pooling, resulting in 4 values per feature map, which is the standard procedure for STL-10 and CIFAR-10 [26, 10, 27, 12]; 2) 3-layer spatial pyramid, i.”
Unsupervised Embedding Learning via Invariant and Spreading Instance Feature (2019, background): “The performances of some state-of-the-art unsupervised methods (k-MeansNet [5], HMP [2], Satck [54] and Exemplar [8]) are also reported.”
Scattering Networks for Hybrid Representation Learning (2018, neutral): “Exemplar CNN [19] 75.4 ± 0.3 Stacked what-where AE [61] 74.33 Hierarchical Matching Pursuit (HMP) [7] 64.5 1 Convolutional K-means Network [13] 60.1 1 We ﬁnd that our model is better in all cases even compared to those utilizing the large unsupervised part of the dataset.”
Stacked What-Where Auto-encoders (2015, neutral): “32% HMP [1] 64 .”
Learning Rich Features from RGB-D Images for Object Detection and Segmentation (2014, background): “[4] look at object detection in RGB-D images, but detect small prop-like objects imaged in controlled lab settings.”
RGB-D datasets using microsoft kinect or similar sensors: a survey (2016, methodology): “This device equipped with intelligent algorithms is contributing to various applications, such as 3D-simultaneous localization and mapping (SLAM) [39, 54], people tracking [69], object recognition [11] and human activity analysis [13, 57], etc.”
Deep Sliding Shapes for Amodal 3D Object Detection in RGB-D Images (2015, methodology): “2D Object Detector in RGB-D Images 2D object detection approaches for RGB-D images treat depth as extra channel(s) appended to the color images, using handcrafted features [9], sparse coding [2, 3], or recursive neural networks [23].”
Convolutional Kernel Networks (2014, neutral): “Method [12] [27] [18] [13] [4] [17] [31] CKN-GM CKN-PM CKN-CO CIFAR-10 82.”
Deep learning for detecting robotic grasps (2013, methodology): “These are computed after the image is aligned to the gripper so that they are always relative to the gripper plates.”
Orientation-boosted Voxel Nets for 3D Object Recognition (2016, background): “Feature learning for 3D recognition has ﬁrst appeared in the context of RGB-D images, where depth is treated as an additional input channel [1, 6, 8].”
Scaling the Scattering Transform: Deep Hybrid Networks (2017, neutral): “Exemplar CNN [11] 75.4 ± 0.3 Stacked what-where AE [43] 74.33 Hierarchical Matching Pursuit (HMP) [3] 64.5 ± 1 Convolutional K-means Network [8] 60.1 ± 1 Table 7.”
An Analysis of Unsupervised Pre-training in Light of Recent Advances (2014, background): “…add it to the hue value for each pixel like so: Convolutional Kernel Networks - Mairal et al. (2014) 62.32 % Hierarchical Matching Pursuit (HMP) - Bo et al. (2013) 64.5 % ± 1.0 % NOMP - Lin and Kung (2014) 67.9 % ± 0.6 % Multi-task Bayesian Optimization - Swersky et al. (2013) 70.1 % ± 0.6 %…”
Multimodal deep learning for robust RGB-D object recognition (2015, methodology): “From left to right: RGB, depth-gray, surface normals [5], HHA [10], our method.”
Learning 6D Object Pose Estimation Using 3D Object Coordinates (2014, background): “More recently a trend has been to learn interest points [22,10], descriptors [30], and matching [14,19,1].”
Ieee Transactions on Pattern Analysis and Machine Intelligence 1 Discriminative Unsupervised Feature Learning with Exemplar Convolutional Neural Networks (2014, neutral): “…1) 4-quadrant max-pooling, resulting in four values per feature map, which is the standard procedure for STL-10 and CIFAR-10 [14], [16], [33], [35] 2) 3-layer spatial pyramid, i.e., max-pooling over the whole image as well as within four quadrants and within the cells of a 4 Â 4 grid,…”
Multiview RGB-D Dataset for Object Instance Detection (2016, methodology): “In addition to the three baselines, we follow the non CNN route of Hierarchical Matching Pursuit (HMP) for the fourth baseline [3].”
Convolutional Clustering for Unsupervised Learning (2015, methodology): “Although unsupervised learning techniques using k-means algorithm were commonly used to train filters in several studies (Coates & Ng, 2011b; Bo et al., 2013), the network encoding structures present many similarities with ConvNets, such as the use of convolution and pooling in each layer.”
Affordance detection of tool parts from geometric features (2015, methodology): “HMP [3] is a hierarchical sparse coding method that learns feature hierarchies called paths.”
Discriminatively Trained Sparse Code Gradients for Contour Detection (2012, background): “Closely related to our work but on the different problem of recognition, Bo et. al. used matching pursuit and K-SVD to learn features in a coding hierarchy [3] and are extending their approach to RGB-D data [4].”