Discovering localized attributes for fine-grained recognition

By Kun Duan, Devi Parikh, David J. Crandall, Kristen Grauman. 2012

302 citations recorded by the source corpus.

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

Joint Discovery of Object States and Manipulation Actions (2017, background): “One solution to recognizing object states would be to manually annotate states for different objects, and treat the problem as a supervised fine-grained object classification task [14, 15].”
Part-Based R-CNNs for Fine-Grained Category Detection (2014, background): “Human-in-the-loop methods [9, 13, 15] ask a human to name attributes of the object, click on certain parts or mark the most discriminative regions to improve classification accuracy.”
Learning Deep Representations of Fine-Grained Visual Descriptions (2016, background): “In particular for the finegrained regime [46, 10, 7, 51], where images of differThe beak is yellow and pointed Accumulate matching score +8.”
Concept Bottleneck Models (2020, methodology): “Similar methods can also be used to refine existing concepts and make them more discriminative (Duan et al., 2012).”
The More You Know: Using Knowledge Graphs for Image Classification (2016, background): “…[8] to vision such as [16] which uses a ﬁxed set of binary attributes to do zero-shot prediction, [33] which uses attributes shared across categories to prevent semantic drift in semi-supervised learning and [5] which automatically discovers attributes and uses them for ﬁne-grained classi-ﬁcation.”
PANDA: Pose Aligned Networks for Deep Attribute Modeling (2013, neutral): “[9] proposed an interactive crowd-sourcing method to discover both localized and discriminative attributes to differentiate bird species.”
Evaluation of output embeddings for fine-grained image classification (2014, background): “Attributes, which are often collected manually [25, 41, 12], have shown promising results in various applications, i.”
P-CNN: Pose-Based CNN Features for Action Recognition (2015, background): “Examples include alignment of faces for face recognition [2] as well as alignment of body parts for recognizing species of birds [12].”
Label-Embedding for Image Classification (2015, neutral): “Several works have also considered the problem of discovering a vocabulary of attributes [5], [16], [36].”
Deformable Part Descriptors for Fine-Grained Recognition and Attribute Prediction (2013, background): “These domains include: dog breed classification [26, 29, 37], subordinate categories of flowers [2, 1, 33, 34] and plants [4, 39], recognition of invertebrates [32], fine-grained classification on subsets of ImageNet [16] such as fungi [15], and species-level categorization of birds [23, 46].”
The Ignorant Led by the Blind: A Hybrid Human–Machine Vision System for Fine-Grained Categorization (2014, methodology): “Interactive methods for generating vocabularies of parts or attributes (Maji 2012; Parikh andGrauman 2011; Duan et al. 2012) and incorporating annotator rationales (Donahue and Grauman 2011), and runtime interactive computer vision systems for segmentation and tracking (Wu and Yang 2006; Rother et a”
Fine-Grained Recognition in the Wild: A Multi-task Domain Adaptation Approach (2017, background): “While fine-grained image recognition is a well studied problem [2, 5, 8, 10, 11, 9, 16, 17, 19, 26], its real world applicability is hampered by limited available data.”
Which Looks Like Which: Exploring Inter-class Relationships in Fine-Grained Visual Categorization (2014, background): “Most of the existing works only focused on solving the first issue by proposing new feature representations, as they emphasized that insufficient discriminative power resides in the way that features are encoded [19,41,18,43,34,40].”
Augmenting Strong Supervision Using Web Data for Fine-Grained Categorization (2015, methodology): “Various methods have been used to define object parts such as unsupervised patch discovery [32, 8], human-in-the-loop methods [12, 10], or direct reliance on strongly supervised datasets with part-level annotations [21, 4, 33, 20].”
Embedding Human Knowledge into Deep Neural Network via Attention Map (2019, methodology): “Although this approach has been widely proposed (Branson et al., 2010; Deng et al., 2013; Bran-son et al., 2011; Parkash and Parikh, 2012; Parikh and Grauman, 2011; Duan et al., 2012), the conventional methods are based on rather small machine learning models comprising fewer model parameters, such ”
Birdsnap: Large-Scale Fine-Grained Visual Categorization of Birds (2014, background): “Much recent work in fine-grained visual categorization has focused on species identification, with work on leaves [14] [25], flowers [19] [25], butterflies [8] [29], insects [17], cats and dogs (e.g., [16] [21]), and birds (e.g., [4] [6] [7] [8] [12] [30] [32] [33]).”