Object recognition with hierarchical kernel descriptors

By Liefeng Bo, Kevin Lai, Xiaofeng Ren, Dieter Fox. 2011

263 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Artificial Intelligence, 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 Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

RGB-D datasets using microsoft kinect or similar sensors: a survey (2016, methodology): “This new dataset further boosts the research on applications such as category recognition, instance recognition, 3D scene labeling and object pose estimation [9, 10, 53].”
Convolutional Kernel Networks (2014, methodology): “The multilayer convolutional kernel slightly differs from the hierarchical kernel descriptors of [2] but exploits similar ideas.”
Joint 2D-3D-Semantic Data for Indoor Scene Understanding (2017, background): “The Computer Vision community has been affected by this change and is experiencing a lot of development in data-driven 2.5D and 3D vision [6, 17, 18, 19, 1].”
Tensor Representations via Kernel Linearization for Action Recognition from 3D Skeletons (2016, methodology): “Similarly, convolutional kernel networks [23] consider stacked layers of a variant of kernel descriptors [22].”
Multimodal deep learning for robust RGB-D object recognition (2015, unknown): “5 HKDES [4] 76.”
Learning hierarchical sparse features for RGB-(D) object recognition (2014, methodology): “By adapting the kernel view to depth maps and 3D points, Bo et al. proposed RGB-D kernel descriptors (2011a; 2011b), and showed that they achieve much higher recognition accuracy than designed feature sets on the RGB-D Object Dataset (Lai et al., 2011a).”
Group Invariance, Stability to Deformations, and Complexity of Deep Convolutional Representations (2017, unknown): “Shallow kernels for images modelling local regions were also proposed in [35], and a multilayer construction was proposed in [6].”
Unsupervised Feature Learning for RGB-D Based Object Recognition (2012, background): “Adapting this view to depth maps and 3D points, RGB-D kernel descriptors are proposed in [4, 6], and the experiments showed that they obtain higher recognition accuracy than hand-designed feature sets on the RGB-D object dataset [19].”
End-to-End Kernel Learning with Supervised Convolutional Kernel Networks (2016, methodology): “Before the resurgence of neural networks, non-parametric models based on positive definite kernels were one of the most dominant topics in machine learning [22].”
Sliding Shapes for 3D Object Detection in Depth Images (2014, background): “3D classification: Classification-based approaches [65–68, 51, 69–79] typically consider the whole object at the same time by extracting a holistic feature for the whole object and classifying the feature vector via a classifier.”
Incremental online learning of objects for robots operating in real environments (2017, unknown): “HKD [24] 76.”
Depth kernel descriptors for object recognition (2011, background): “The normalized linear kernel m̃pm̃q weighs the contribution of each gradient where m̃p = mp/ √ ∑p∈P m2p + εg and εg is a small positive constant to ensure that the denominator is larger than 0; the position Gaussian kernel ks(p,q) = exp(−γs‖p− q‖2) measures how close two pixels are spatially; the…”
Histogram of Oriented Normal Vectors for Object Recognition with a Depth Sensor (2012, methodology): “We present the performance of 6 methods: HOG, HOGD, HONV, Depth Kernel Descriptors (DKD) [7], Hierarchical Kernel Descriptors (HKDES) [8] and Instance Distance Learning (IDL) [9].”
Hand-Crafted Features or Machine Learnt Features? Together They Improve RGB-D Object Recognition (2014, background): “Hierarchical kernel descriptors [2] is proposed to generate image-level features applying kernel descriptors layer by layer.”
3D Semantic Parsing of Large-Scale Indoor Spaces (2016, methodology): “…approach from existing techniques: 1) processing a large-scale point cloud of an entire building (indoor spaces), rather than one or few RGB-D images, 2) detection of space dividers (walls) based on their void (empty) space rather than planar-surface/linear-boundary assumptions, 3) utilizing a…”
Invariance and Stability of Deep Convolutional Representations (2017, unknown): “Shallow kernels for images modelling local regions were also proposed in [25], and a multilayer construction was proposed in [4].”
Bilinear heterogeneous information machine for RGB-D action recognition (2015, background): “Due to the advent of low-cost Kinect sensors [16], lots of attempts have been devoted to object recognition [3, 2] and action recognition [10, 13, 25, 5] from depth images.”