Combining randomization and discrimination for fine-grained image categorization

By Bangpeng Yao, Aditya Khosla, Li Fei-Fei. 2011

336 citations recorded by the source corpus.

Subfields: 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 Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

CNN Features Off-the-Shelf: An Astounding Baseline for Recognition (2014, methodology): “RF [42] uses a bag of decision trees trained on a bag of SIFT words.”
Contextual Action Recognition with R*CNN (2015, background): “[33] densely sample image regions at arbitrary locations and scales with reference to the ground-truth region.”
Actions and Attributes from Wholes and Parts (2014, methodology): “[21] uses a holistic CNN classifier that outperforms partbased approaches [20, 30].”
Robust 3D Action Recognition with Random Occupancy Patterns (2012, background): “Randomization has been applied in [8] and [9] to address this problem.”
Leveraging the Wisdom of the Crowd for Fine-Grained Recognition (2016, methodology): “All images are cropped to the bounding boxes, as is standard for many published results [15], [16], [18], [35], [43].”
TriCoS: A Tri-level Class-Discriminative Co-segmentation Method for Image Classification (2012, background): “Some of the previously published results make use of those hand annotations both during training and testing time [25,26].”
Deformable Part Descriptors for Fine-Grained Recognition and Attribute Prediction (2013, methodology): “We also compare with other published methods, including MKL [13], random forest [46], TriCos [14], template matching [43], segmentation [1] and the recently-published Bubblebank method [17].”
Hidden stratification causes clinically meaningful failures in machine learning for medical imaging (2019, background): “Problems similar to hidden stratification have been observed or postulated in many domains, including traditional computer vision [30], fine-grained image recognition [39], genomics [7], and epidemiology (often termed “spectrum effects”) [23].”
Discovering localized attributes for fine-grained recognition (2012, background): “[2, 3, 6, 9, 11, 14, 28]).”
Dog Breed Classification Using Part Localization (2012, methodology): “In addressing this problem, [14] mines discriminative features with randomized sampling, and [3] uses a multiple kernel framework to combine kernels in a way that is most discriminative for each class.”
Which Looks Like Which: Exploring Inter-class Relationships in Fine-Grained Visual Categorization (2014, methodology): “We compare with a multiple kernel learning (MKL) method [11], a bag-of-features approach using the LLC [36], a randomization based method [42], a multi-cue representation [26],”
Action Recognition with Exemplar Based 2.5D Graph Matching (2012, background): “However, while such approaches sound promising, the winning method [9] in the recent PASCAL challenge [10] simply treats action recognition as an image classification problem, without explicitly modeling human poses.”
On-Board Object Detection: Multicue, Multimodal, and Multiview Random Forest of Local Experts (2017, methodology): “The method in [33] also makes use of RF with local classifiers at the node level, although it requires to extract many complex region-based descriptors, making it computationally more demanding than [27].”
Incremental Learning of Random Forests for Large-Scale Image Classification (2016, methodology): “Such a system is described in [12], where randomness comes from window extraction in the images.”
Novel Dataset for Fine-Grained Image Categorization : Stanford Dogs (2012, methodology): “We introduce a 120 class Stanford Dogs dataset, a challenging and large-scale dataset aimed at fine-grained image categorization.”