Harvesting Image Databases from the Web

By Florian Schroff, Antonio Criminisi, Andrew Zisserman. 2007

434 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Molecular Biology

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

In Molecular Biology

Citations in the corpus

Learning Video Representations from Textual Web Supervision (2020, methodology): “The most commonly-used approach is to use image search results, and to label each image with the query that was used to ﬁnd it [65, 15, 56, 6, 11, 13, 21, 10, 18, 35].”
Webly Supervised Learning of Convolutional Networks (2015, background): “In fact, they are so clean that they are biased towards iconic images where a single object is centered with a clean background in a canonical viewpoint [31, 40, 3, 30].”
The Unreasonable Effectiveness of Noisy Data for Fine-Grained Recognition (2015, methodology): “4 is inspired by methods that propose to learn from web data [13, 8, 9, 60, 40, 29, 16] or reason about label noise [34, 57, 52, 47, 38].”
Large-Scale Weakly-Supervised Pre-Training for Video Action Recognition (2019, background): “Dataset sources: Many prior approaches use Internet as a natural source of content [6, 12, 13, 18, 24, 25, 34, 37, 44, 46, 51, 58, 67] and the associated search queries, hashtags, or user-comments as labels to construct datasets.”
Domain Adaptation for Visual Applications: A Comprehensive Survey (2017, methodology): “es [3, 121], semantic [246, 174, 87, 256] or multi-view embedding [110, 153, 196, 104, 25, 238] especially used to perform cross-modal image retrieval. In the same spirit, webly supervised approaches [82, 242, 189, 14, 62, 37, 93] are also related to DA and HDA as is these approaches rely on collect”
Keywords to visual categories: Multiple-instance learning forweakly supervised object categorization (2008, methodology): “We demonstrate our method with benchmark datasets, and show that it performs well relative to both state-of-the-art unsupervised approaches and traditional fully supervised techniques.”
Landmark classification in large-scale image collections (2009, background): “, [2, 14, 21]), however such approaches generally make use of image features and machine learning techniques, and thus the resulting datasets are not independent of the kinds of features and methods that one wants to test.”
Harvesting large-scale weakly-tagged image databases from the web (2010, methodology): “Some pioneering works have been done to leverage Internet images for computer vision tasks [2, 4-8].”
Baby talk: Understanding and generating simple image descriptions (2011, background): “Other object classes that have been considered include animal images from the web [7], [30], [37] where text from the containing webpage can be utilized for improved image classification.”
Unsupervised Learning of Visual Sense Models for Polysemous Words (2008, result): “Category classification experiments show that our dictionarybased approach outperforms baseline methods.”
An Attribute-Assisted Reranking Model for Web Image Search (2015, background): “[3] learned a query independent text based re-ranker.”
Learning Models for Object Recognition from Natural Language Descriptions (2009, methodology): “searching Google Images for “car” or “voiture” to find example car images [10, 17, 19], in this paper we consider the use of Natural Language Processing (NLP) methods to extract information beyond keywords from naturally occurring English text.”
WebVision Database: Visual Learning and Understanding from Web Data (2017, background): “Meta Information: One advantage of web images is the abundant textual information, which usually contains valuable semantic information about the images, and has been shown to be quite useful for image categorization in the literature[30, 22, 15].”
Learning Everything about Anything: Webly-Supervised Visual Concept Learning (2014, background): “Learning from web images: Due to the complexity of the detection task and the higher supervision requirements, most previous works [4, 19, 28, 38, 44, 48] on using web images have focused on learning models only for image classification.”
Learning Object Categories From Internet Image Searches (2010, methodology): “Schroff et al. [34] extend this method to be fully automatic by first ranking the images based on the text and other metadata, and then learning a visual model discriminatively from the highly ranked images by using a support vector machine (SVM) classifier to cope with the noisy image labels.”
Well Begun Is Half Done: Generating High-Quality Seeds for Automatic Image Dataset Construction from Web (2014, methodology): “But previous works mainly focus on the second step, paying relative less attention to the first step – the seeds are either labeled by human [10,14,13,5] (high cost) or labeled according to top-ranked images [14,13] or noisy surrounding text [1,16] (low quality).”