Word Spotting in the Wild

By Kai Wang, Serge J. Belongie. 2010

531 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

Towards VQA Models That Can Read (2019, background): “Several existing datasets study text detection and/or parsing in natural everyday scenes: COCO-Text [43], Street-View text [44] IIIT-5k [33] and ICDAR 2015 [22].”
Synthetic Data for Text Localisation in Natural Images (2016, methodology): “ICDAR [23, 24, 39], the Street View Text (SVT) dataset [43].”
COCO-Text: Dataset and Benchmark for Text Detection and Recognition in Natural Images (2016, background): “tated instances per image for COCO-Text, ICDAR03 and ICDAR15. (d) Distribution of instance sizes for same datasets. (e) Spacial text occurrence probability for same datasets. 2 ther, Street View Text [20] contains a total of 350 images from Google Street View and 725 total labeled words, however it ”
End-to-End Text Recognition with Hybrid HMM Maxout Models (2013, methodology): “We test our word recognition subsystem on word images from the ICDAR 2003 (Lucas et al., 2003) and SVT (Wang & Belongie, 2010) word recognition test sets.”
Scene Text Visual Question Answering (2019, neutral): “For the strongly contextualised task (1), following the standard practice used for end-to-end word spotting [27, 26, 55], we create a dictionary per image that contains the words that appear in the answers defined for questions on that image, along with a series of distractors.”
Focusing Attention: Towards Accurate Text Recognition in Natural Images (2017, background): “Early works mainly develop bottom-up approaches: first detecting characters one by one by sliding window [29, 30], connected components [21] or Hough voting [32], then integrating these characters into the output text.”
Robust Seed Localization and Growing with Deep Convolutional Features for Scene Text Detection (2015, methodology): “Robust Reading Competition datasets of ICDAR 2011 [15], ICDAR 2013 (Challenge 2: Reading Text in Scene Images) dataset [5] and Street View Text (SVT) [17] dataset.”
Supervised mid-level features for word image representation (2014, background): “The most popular approaches to address this task involve detecting and localizing individual characters in the word image and using that information to infer the contents of the word, using for example conditional random fields and language priors [39, 19, 20, 5, 22].”
A Unified Framework for Multioriented Text Detection and Recognition (2014, methodology): “Most of the existing methods for text detection [6], [10], [11], [26] and recognition [7], [8], [18] can only handle horizontal or near-horizontal texts.”
TextProposals: A text-specific selective search algorithm for word spotting in the wild (2016, methodology): “We make use of the following scene text datasets: the different datasets from the ICDAR Robust Reading Competitions series (ICDAR2003 [57], ICDAR2013 [58], and ICDAR2015 [4]), the Street View Text dataset (SVT) [14], and the Multi-Language end-to-end (MLe2e) dataset [59, 60].”
Cascaded Segmentation-Detection Networks for Text-Based Traffic Sign Detection (2018, unknown): “Wang and Belongie [33] first use”
End-to-end scene text recognition (2011, methodology): “In this benchmark, the organizers identified four subproblems: (1) cropped character classification, (2) full image text detection, (3) cropped word recognition, and (4) full image word recognition.”
Detecting texts of arbitrary orientations in natural images (2012, background): “There have been a large number of methods dealing with text detection in natural images and videos [18, 16, 30, 27, 22, 26].”
Recognizing Text with Perspective Distortion in Natural Scenes (2013, background): “It has been shown that scene characters can be extracted as MSERs [11, 12].”
ESIR: End-To-End Scene Text Recognition via Iterative Image Rectification (2018, methodology): “Different scene character detection and recognition methods have been reported by using sliding window [43, 42], connected components [31], extremal regions [32], Hough voting [47], co-occurrence histograms [41], etc.”
Symmetry-Constrained Rectification Network for Scene Text Recognition (2019, methodology): “A popular pipeline of traditional methods [10, 50, 49, 51, 38, 37, 39, 55, 4] is in the bottom-up architecture.”
Large-Lexicon Attribute-Consistent Text Recognition in Natural Images (2012, methodology): “Alternative approaches (which could also be used within our framework) include using connected components resulting from stroke transforms of edge maps [5], or using sliding HOG-templates [1].”
Top-down and bottom-up cues for scene text recognition (2012, methodology): “Here we show a few sample images from the SVT [30] and ICDAR [1] datasets, with issues such as, very different fonts, shadows, low resolution, occlusions.”