Consumer video understanding: a benchmark database and an evaluation of human and machine performance

By Yu-Gang Jiang, Guangnan Ye, Shih-Fu Chang, D. Ellis, Alexander C. Loui. 2011

359 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

Actions ~ Transformations (2015, background): “Recently, a few large-scale video datasets have been introduced (e.g. CCV [20], Sports-1M [21], ActivityNet [10], THUMOS [19] and FGA-240 datasets [45]).”
Modeling Spatial-Temporal Clues in a Hybrid Deep Learning Framework for Video Classification (2015, methodology): “We follow the convention deﬁned in [17] to use a training set of 4,659 videos and a test set of 4,658 videos.”
Semantic Pooling for Complex Event Analysis in Untrimmed Videos (2016, background): “CCV sub : The ofﬁcial Columbia Consumer Video dataset [70] contains 9,317 videos in total with 20 semantic categories, including scenes like “beach,” objects like “cat,” and events like “baseball” and “parade.””
High-level event recognition in unconstrained videos (2012, methodology): “Since the early concatenation of features may amplify the “curse of dimensionality” problem, late fusion has been frequently adopted for multimodal event recognition in unconstrained videos [48,57,58,96].”
Fusing Multi-Stream Deep Networks for Video Classification (2015, methodology): “Following [20], we adopt the suggested training and test split and compute the average precision (AP) for each class.”
Video Event Detection by Inferring Temporal Instance Labels (2014, methodology): “In terms of selecting multi-granular instances, we evaluated combinations of various instances on CCV dataset.”
Attribute Learning for Understanding Unstructured Social Activity (2012, methodology): “The SIFT, STIP and MFCC features for all these videos are extracted according to [16], and included in the dataset.”
Robust Multi-View Spectral Clustering via Low-Rank and Sparse Decomposition (2014, methodology): “Following the settings in (Kumar , Rai, and Daumé 2011), we use the Gaussian kernel for each view, and the best clustering results are reported with the parameter λ being chosen from 0.01 to 0.05. dataset instances views clusters BBC 2225 3 5 BBCSport 737 2 5 WebKB 1051 2 2 UCI Digits 2000 3 10 Flow”
Better and Faster: Knowledge Transfer from Multiple Self-supervised Learning Tasks via Graph Distillation for Video Classification (2018, methodology): “, 2013] and CCV [Jiang et al., 2011], which contain 13k, 6.”
A Review of Human Activity Recognition Methods (2015, background): “Lan et al. (2012b) recognized group activities, whichwere considered as latent variables, encoding the contextual information in a video sequence.”
An Adaptive Semisupervised Feature Analysis for Video Semantic Recognition (2018, methodology): “In the experiments, we directly use the space-time interest points (STIPs) features with 5000-D bag-of-visual-words (BoWs) to represent the low-level feature of videos [42].”
Multi-level Feature Learning for Contrastive Multi-view Clustering (2021, background): “Columbia Consumer Video (CCV) [15] is a video dataset with 6,773 samples belonging to 20 classes and provides hand-crafted Bag-of-Words representations of three views, such as STIP, SIFT, and MFCC.”
Exploiting Feature and Class Relationships in Video Categorization with Regularized Deep Neural Networks (2015, methodology): “The CCV dataset was constructed in 2011, aiming to stimulate the research on Internet consumer video analysis [54].”