ICME 2017

170 papers
The 2017 IEEE International Conference on Multimedia and Expo (ICME) featured 245 papers, with a primary research focus on structural similarity for perceptual audio modeling. The technical program highlighted advancements in visual tracking, including the development of multi-task dense networks and attentional recurrent neural networks. Video analysis remained a central theme, with studies addressing salient object detection through cross-frame cellular automata and 3-D padding techniques for 360-degree video projection. Other research directions included kinship verification using status-aware projection metric learning, image-based user profiling, and reflection removal via sparsity-based patch searching. Prolific authors contributing to this year's proceedings included Chen Chen, Qingming Huang, Shiqi Wang, and Guangtao Zhai. The collection also featured work on interactive systems for low-poly illustration generation and adaptive thinning. These contributions reflect a diverse range of interests spanning audio-visual modeling, computer vision, and specialized image processing applications.

Top Research Directions

Research directions most represented by ICME 2017 papers.

Representational Interference in Feature Hashing 1
Aligning Confidence and Localization Quality in Object Detectors 1
Minimizing Intra-Class Variance with Center Loss 1
Learnable Upsampling for Dense Spatial Prediction 1
Dynamic Scale Adaptation for Visual Object Tracking 1
Joint Classification and Metric Learning for Robust Embeddings 1
Modeling Narratives and Sequences as Directed Temporal Graphs 1
Stochastic Temporal Sampling for Motion Robustness 1
Deep Metric Learning via Triplet Constraints 1
Efficient Feature and Search Management for Real-Time Tracking 1
The Ground Truth Dependency in Visual Quality Assessment 1
Lexical Identity Bias in Automated Content Moderation 1
Modeling Intra-Class Variance with Multi-Prototype Representations 1
Factorizing Speech into Content Timbre and Prosody 1
Discriminating Categories with Low Inter-Class Variance 1

Top Authors

AuthorPapers
Chen Chen4
Fumin Shen4
Mai Xu4
Qingming Huang4
Shiqi Wang4
Guangtao Zhai3
Heng Tao Shen3
Hong Liu3
Hongtao Lu3
Huabing Zhou3
Jiayi Ma3
Jingyu Yang3
Jun Chen3
Ke Lv3
Risheng Liu3
Xin Fan3
Xing Xu3
Yang Yang3
Yao Zhao3
Boxin Shi2

Impactful Papers

Papers from ICME 2017 that shaped research clusters.

Foundational

Influential

Papers

Title
Steganographer detection via deep residual network
Supervised cross-modal hashing without relaxation
SynCam: Capturing sub-frame synchronous media using smartphones
Temporal aggregation for first-person action recognition using Hilbert-Huang transform
Temporally enhanced image object proposals for videos
The OUC-vision large-scale underwater image database
Time-ordered spatial-temporal interest points for human action classification
Top attention in line with time: A light-weight strategy
Topology preserving graph matching for partial face recognition
Towards sketch-based image retrieval with deep cross-modal correlation learning
Universal enhancement of salient object detection
Unsupervised cross-modal retrieval through adversarial learning
Unsupervised feature selection with ordinal locality
Video salient object detection via cross-frame cellular automata
VIDEOWHISPER: Towards unsupervised learning of discriminative features of videos with RNN
Vision-based parking-slot detection: A benchmark and a learning-based approach
Visual relationship detection with object spatial distribution
Weakly structured information aggregation for upper-body posture assessment using ConvNets
Webly-supervised visual concept learning with cardinality guided instance mining and clustered multitask refinement
Weighted sequence loss based spatial-temporal deep learning framework for human body orientation estimation

« prev Page 4/4