HICO: A Benchmark for Recognizing Human-Object Interactions in Images

By Yu-Wei Chao, Zhan Wang, Yugeng He, Jiaxuan Wang, Jia Deng. IEEE International Conference on Computer Vision, 2015

339 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/c0fa318f20ed622a3f28e424b8d98ff5f8ca41b5

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

Neural Motifs: Scene Graph Parsing with Global Context (2017, background): “Actions and relations have been a particularly fruitful source of context [27, 45], especially when combined with pose to create human-object interactions [43, 3].”
You2Me: Inferring Body Pose in Egocentric Video via First and Second Person Interactions (2019, background): “Beyond body pose, there is a growing interest in modeling human-human interactions [57, 23, 40] to predict pedestrian trajectories [1, 43, 2], analyze social behavior and group activities [43, 59, 6, 15, 23], and understand human-object interactions [61, 19, 12].”
Learning Prototypes for Visual Relationship Detection (2018, background): “Concurrently, datasets providing object bounding box annotations and relationships for pairs of objects have been released, allowing to train visual relationship detection models [6], [12], [18] and to test them on seen, unseen [12] or unusual relationships [16].”
Crowdsourcing in Computer Vision (2016, positive): “Chao et al. [2015] used Mechanical Turk to create the Humans Interacting with Common Objects (HICO) dataset, containing 47,774 images covering 600 categories of human-object interactions (e.g., “ride a bike”) over 117 common actions (e.g., “ride,” “feed”) performed on 80 common objects (e.g., “bike,”
AVA: A Video Dataset of Spatio-Temporally Localized Atomic Visual Actions (2017, background): “The AVA dataset is also related to still image action recognition datasets [7, 9, 12].”
Under review as a conference paper at ICLR 2017 D ISCOVERING OBJECTS AND THEIR RELATIONS FROM ENTANGLED SCENE REPRESENTATIONS (2016, neutral): “…to be useful for solving tasks that require reasoning not only about object-object relations, but also about verb- object relations, as in human-object interaction datasets (Chao et al., 2015) or question-answering tasks that involve reasoning between multiple objects (Krishna et al., 2016).”
Reconstructing Hands in 3D with Transformers (2023, methodology): “Among the larger scale efforts, COCO-WholeBody [25] provides hand annotations for the people in the COCO dataset [35] and Halpe [15] annotates hands in the HICO-DET dataset [4, 5].”
Unsupervised Visual Sense Disambiguation for Verbs using Multimodal Embeddings (2016, neutral): “HICO doesn’t aim for complete coverage, but re-stricts itself to the top three WordNet senses of a verb.”
MECCANO: A Multimodal Egocentric Dataset for Humans Behavior Understanding in the Industrial-like Domain (2022, background): “HICO-Det [7] is a large-scale dataset composed of static images used as a benchmark to study the task of HOI detection.”
Who is Mistaken? (2016, neutral): “[5, 28, 7, 20, 10].”
Detecting and Recognizing Human-Object Interactions (2017, background): “The older TUHOI [19] and HICO [4] datasets only have imagelevel labels and thus do not allow for grounding interactions in a detection setting, while COCO-a [28] is promising but only a small beta-version is currently available.”
Learning Models for Actions and Person-Object Interactions with Transfer to Question Answering (2016, methodology): “We train deep networks on the HICO [14] and MPII [15] datasets to predict human activity.”
Learning Human-Object Interactions by Graph Parsing Neural Networks (2018, background): “More recently, inspired by the notable success of deep learning and the availability of large-scale HOI datasets [1,2], several deep learning based HOI models were then proposed.”
A Comprehensive Survey of Vision-Based Human Action Recognition Methods (2019, background): “[27] released the HOIC dataset in 2015, including more than 600 interactions across 47,774 images.”
Learning to Detect Human-Object Interactions (2017, methodology): “As noted in [1], a person can concurrently perform different classes of actions to a target object, e.”
A Comprehensive Survey of Scene Graphs: Generation and Application (2021, methodology): “For HOI, there are two main benchmarks: HICO-DET [179] and HCVRD [150].”
Webly Supervised Concept Expansion for General Purpose Vision Models (2022, background): “We source verbs from a combination of vision datasets with large verb vocabularies including imSitu [85], HICO [9] and VRD [48].”
The MECCANO Dataset: Understanding Human-Object Interactions from Egocentric Videos in an Industrial-like Domain (2020, background): “Among the most notable datasets, we can mention V-COCO [31], which adds 26 verb labels to the 80 objects of COCO [40], HICO-Det [8], labeled with 117 verbs and 80 objects, HOIA [38], which focuses on 10 verbs and 11 objects indicat-”