Introduction to a Large-Scale General Purpose Ground Truth Database: Methodology, Annotation Tool and Benchmarks

By Benjamin Z. Yao, Xiong Yang, Song-Chun Zhu. 2007

246 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Evaluation & Benchmarks

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 Evaluation & Benchmarks

Citations in the corpus

ImageNet Large Scale Visual Recognition Challenge (2014, background): “The LotusHill dataset (Yao et al., 2007) contains very detailed annotations of objects in 636,748 images and video frames, but it is not available for free.”
Do ImageNet Classifiers Generalize to ImageNet? (2019, neutral): “[60] Benjamin Z.”
OPTIMOL: Automatic Online Picture Collection via Incremental Model Learning (2007, methodology): “Among the solutions, one of the major trends is to manually collect and annotate a ground truth dataset (LotusHill ( Yao et al. 2007 ) and LableMe (Russell et al. 2005 )). Due to the vast number of object classes in our world, however, manually collecting images for all the classes is currently impo”
I2T: Image Parsing to Text Description (2010, background): “As stated in a report [42], this data set provides ground truth annotation for a range of vision tasks from high-level scene classification and object Fig.”
Image retrieval using scene graphs (2015, result): “While this level of detail is similar in spirit to the Lotus Hill dataset [67], our dataset is based on an open universe assumption and is freely available.”
A visual Turing test for computer vision systems (2014, methodology): “Popular datasets for this task include the Pascal dataset [4], the LabelMe dataset [24], and the Lotus Hill dataset [25], all populated by relatively unconstrained natural images, but varying considerably in size and in the level of annotation, ranging from a few keywords to hierarchical…”
Semantic object classes in video: A high-definition ground truth database (2009, background): “Imageparsing.com (Yao et al., 2007) has unknown numbers of its many classes, and is said to still be growing.”
ImageNet: A large-scale hierarchical image database (2009, methodology): “LabelMe and Lotus Hill datasets LabelMe [21] and the Lotus Hill dataset [27] provide 30k and 50k labeled and segmented images, respectively 5.”
Evaluating information contributions of bottom-up and top-down processes (2009, methodology): “The data. Here, we need the data with ground truth for both objects and their parts, and we found the data from LHI database [ 17 ] which fits the requirement.”
Crowdsourcing in Computer Vision (2016, unknown): “The proprietary LotusHill dataset [Yao et al., 2007] contains detailed annotations of objects in 636,748 images and video frames.”
Quality Assessment for Crowdsourced Object Annotations (2011, background): “Whether such workers are paid [9] or not [4], individual annotations can not be trusted as they can for a dataset such as Pascal or Lotus Hill [2].”
What, where and who? Classifying events by scene and object recognition (2007, methodology): “We have also obtained a thorough groundtruth annotation for every image in the dataset (in collaboration with Lotus Hill Research Institute [26]).”
The segmented and annotated IAPR TC-12 benchmark (2010, background): “[23] have also been putting tremendous effort into creating another large-scale benchmark collection.”
Poselets: Body part detectors trained using 3D human pose annotations (2009, background): “tations [16, 21 ] and 2D joint locations [7, 21, 19].”