The segmented and annotated IAPR TC-12 benchmark

By Hugo Jair Escalante, Carlos A. Hernández, Jesus A. Gonzalez, A. López-López, Manuel Montes-y-Gómez, Eduardo F. Morales, Luis Enrique Sucar, Luis Villaseñor-Pineda, Michael Grubinger. Computer Vision and Image Understanding, 2010

422 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/50a331a2adb66ecfd98052ede37be1799a671585

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

Natural Language Object Retrieval (2015, methodology): “It contains 20,000 images from IAPR TC-12 dataset [9], together with segmented image regions from SAIAPR-12 dataset [5], and 120K annotated descriptions for the image regions collected in a two-player game that aims to make the image region identifiable from the annotation.”
High-Dimensional Sparse Cross-Modal Hashing with Fine-Grained Similarity Embedding (2021, neutral): “To evaluate the performance of HSCH, we conducted extensive experiments on three widely-used datasets, i.e., MIRFlickr-25K [17], IAPR TC-12 [11], and NUS-WIDE [6].”
Referring Expression Comprehension via Cross-Level Multi-Modal Fusion (2022, methodology): “RefCOCO, RefCOCO+, and RefCOCOg are collected from MS-COCO [28], whereas ReferItGame is collected from SAIAPR-12 [12].”
Referring Image Segmentation via Recurrent Refinement Networks (2018, methodology): “It is built upon IAPR TC-12 dataset [3] and consists of both objects and stuff, e.”
ReferItGame: Referring to Objects in Photographs of Natural Scenes (2014, background): “Crucial for our purposes, the SAIAPR TC-12 expansion (Escalante et al., 2010) includes segmentations of each image into regions indicating the locations of constituent objects.”
Understanding and predicting importance in images (2012, positive): “Crucially, in its expansion (SAIAPR TC-12), each image is also segmented into constituent objects and labeled according to a set of (275) labels [5].”
Improving One-stage Visual Grounding by Recursive Sub-query Construction (2020, methodology): “The ReferItGame dataset [17] has 20,000 images from SAIAPR-12 [11].”
Generation and Comprehension of Unambiguous Object Descriptions (2015, background): “This leads to different styles of annotations. ing box), and it must generate a referring expression for the target object.”
Deep Cross-Modal Hashing (2016, methodology): “The IAPR TC-12 dataset [8] consists of 20,000 imagetext pairs which are annotated using 255 labels.”
Cross-Modal Self-Attention Network for Referring Image Segmentation (2019, methodology): “The ReferIt dataset is built upon the IAPR TC-12 [6] dataset.”
Entity-Enhanced Adaptive Reconstruction Network for Weakly Supervised Referring Expression Grounding (2022, positive): “It contains 20,000 annotated images from IAPR TC-12 dataset [56] and SAIAPR-12 dataset [57].”
A Fast and Accurate One-Stage Approach to Visual Grounding (2019, methodology): “ReferItGame [15] has 20,000 images from the SAIAPR-12 dataset [6].”
Segmentation from Natural Language Expressions (2016, methodology): “In this dataset, the ground-truth segmentation comes from the SAIAPR-12 dataset [28].”