CVPR 2018

1,039 papers

Top Research Directions

The 2,042 research directions with CVPR 2018 papers, ranked by paper count, next to 2017. Share is the fraction of the edition’s papers in the direction.

#Direction20182017Share 2018Share 20172017 rank
1Generalizing from Synthetic Data to the Real World24112.3%1.3%4
2Deep Learning on Unstructured 3D Point Clouds1811.7%0.1%673
3Preserving Tiny Features Against Spatial Resolution Degradation18141.7%1.7%1
4High-Resolution Generative Scaling1631.5%0.4%131
5Adversarial Training for Domain Invariant Representations1511.4%0.1%508
6Exploiting Temporal Redundancy in Video Neural Networks14121.3%1.5%2
7Bridging the Weakly Supervised Localization Gap1351.3%0.6%27
8Exploiting Spatial Sparsity in Volumetric Neural Networks1351.3%0.6%32
9Composite Loss Functions for Perceptual Image Restoration1281.2%1.0%8
10Eliminating the Optical Flow Bottleneck in Video Understanding1291.2%1.1%6
11Test Time Adaptation for Distribution Shift1131.1%0.4%194
12Hierarchical Temporal Modeling for Long Video Understanding1051.0%0.6%35
13Resolving Depth Ambiguity in Monocular 3D Reconstruction10121.0%1.5%3
14Dense Prediction Adaptation930.9%0.4%106
15Generative Hallucination for Disoccluded Region Synthesis930.9%0.4%127
16Learning Depth Estimation without Dense Ground Truth910.9%0.1%964
17Expanding the Receptive Field of Convolutional Networks860.8%0.7%17
18Harnessing Parameter Redundancy in Deep Neural Networks810.8%0.1%849
19Learning Object Localization from Image Level Labels820.8%0.2%333
20Resolving the Ambiguity of Inverse Rendering840.8%0.5%83
21The Inference Latency of Iterative Refinement820.8%0.2%465
223D Convolutions for Spatiotemporal Feature Learning740.7%0.5%50
23Achieving Pose Invariance in Deep Facial Analysis710.7%0.1%492
24Data Augmentation in the Latent Space710.7%0.1%655
25Disentangling Attribute Control in Generative Models720.7%0.2%269

Fastest growing

Largest gain in share since 2017, among directions with at least 5 papers in 2018.

Direction20172018Share change
Deep Learning on Unstructured 3D Point Clouds118×14.3
Adversarial Training for Domain Invariant Representations115×11.9
Learning Depth Estimation without Dense Ground Truth19×7.2
Harnessing Parameter Redundancy in Deep Neural Networks18×6.4
Achieving Pose Invariance in Deep Facial Analysis17×5.6
Data Augmentation in the Latent Space17×5.6
Cascaded Error Propagation in 2D to 3D Lifting16×4.8
Compressing Deep Networks for Edge Hardware Constraints16×4.8
Dense Feature Reuse in Deep Neural Networks16×4.8
End-to-End Representation Learning for Video Understanding16×4.8

New this year

Directions with no 2017 paper and at least 5 in 2018.

Direction2018
Differentiable Coordinate Regression with Soft Argmax6
Coarse to Fine Refinement for Motion Estimation5
Hierarchical Feature Alignment for Domain Adaptation5
Scaling Facial Action Recognition Without Expert Labels5
Stabilizing Adversarial Networks with Wasserstein Distance and Gradient Penalties5
Zero-Shot Generalization through Diverse Synthetic Supervision5

Lost share

Largest loss in share since 2017, among directions with at least 5 papers in 2017.

Direction20172018Share change
Jointly Optimizing Representations and Clusters81×0.1
Overcoming the Manual Attribute Annotation Bottleneck61×0.1
Spatial Context Preservation in Multiple Instance Learning61×0.1
Accelerating Neural Networks via Frequency Domain Convolutions51×0.2
Determinantal Point Processes for Diverse Subset Selection51×0.2
Generalizing Emotion Recognition to Unconstrained Environments51×0.2
Modeling the Temporal Evolution of Facial Expressions51×0.2
Segmenting Disconnected Object Fragments Under Occlusion51×0.2
Uncalibrated Photometric Stereo82×0.2
Direct Discrete Optimization for Binary Representation Learning72×0.2

Download all directions as CSV

Institutions

Institutions credited by CVPR 2018 papers: a paper credits every institution named in its affiliations, once. 507 institutions share 1,026 credited papers.

#InstitutionPapersShare20172017 rank
1Google (United States) 464.5%277
2Chinese Academy of Sciences 444.3%333
3Carnegie Mellon University 403.9%421
4Tsinghua University 393.8%296
5Meta (United States) 383.7%1814
6Sensetime (China) 353.4%1229
7University of Chinese Academy of Sciences 353.4%314
8Chinese University of Hong Kong 343.3%1913
9ETH Zurich 313.0%2310
10Adobe Systems (United States) 302.9%259
11Microsoft (United States) 302.9%402
12University of California, Berkeley 292.8%258
13Stanford University 282.7%305
14Tencent (China) 282.7%851
15Nanyang Technological University 242.3%1323
16Peking University 242.3%764
17Computer Vision Laboratory 222.1%858
18Massachusetts Institute of Technology 222.1%1326
19National University of Singapore 222.1%1420
20University of Oxford 222.1%2012
21Adelaide University 201.9%1518
22Shanghai Jiao Tong University 201.9%761
23University of Southern California 181.8%768
24Australian National University 171.7%1616
25Nvidia (United States) 171.7%3159
26State Key Laboratory of Pattern Recognition 171.7%1421
27University of California, Merced 171.7%760
28University of Science and Technology of China 171.7%1134
29University of Toronto 171.7%945
30Alibaba Group (China) 161.6%587

Download all institutions as CSV

About this data

Papers. CVPR 2018 lists 1,039 papers on Lacuna, duplicate versions counted once; rejected and withdrawn submissions are left out.

Directions. Directions are Lacuna’s clustering of the concept elements extracted from each paper, so a paper can belong to several and direction counts add up to more than the paper count. A direction’s share is its papers divided by the edition’s papers. Comparisons with 2017 use shares, since the edition went from 827 to 1,039 papers.

Institutions. A paper credits each institution named in its affiliations, resolved to a ROR record, once. 1,026 of 1,039 papers (98.7%) credit at least one institution and are the base for institution shares; the rest have no resolved affiliation yet.

Data. Counts come from Lacuna’s data build of 2026-11-10 and change as papers are added. Tables as CSV: directions · institutions. Cite as: Lacuna, “CVPR 2018”, https://lacuna.tiptreesystems.com/venue/cvpr/2018.

Top Authors

AuthorPapers
Ming-Hsuan Yang17
Luc Van Gool16
Wei Liu12
Xiaogang Wang12
Jan Kautz11
Chunhua Shen10
Jiashi Feng10
Shuicheng YAN10
Bernt Schiele9
Huchuan Lu9
Junjie Yan9
Liang Lin9
Raquel Urtasun9
Tao Xiang9
Timothy Hospedales9
Wanli Ouyang9
Huchuan Lu8
Serge Belongie8
Song-Chun Zhu8
Gang Wang7

Impactful Papers

Papers from CVPR 2018 that shaped research clusters.

Foundational

Influential

Papers

Title
2018 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2018, Salt Lake City, UT, USA, June 18-22, 2018
2018 IEEE Conference on Computer Vision and Pattern Recognition Workshops, CVPR Workshops 2018, Salt Lake City, UT, USA, June 18-22, 2018
2D/3D Pose Estimation and Action Recognition Using Multitask Deep Learning
3D Human Pose Estimation in the Wild by Adversarial Learning
3D Human Pose Estimation in Video With Temporal Convolutions and Semi-Supervised Training
3D Human Sensing, Action and Emotion Recognition in Robot Assisted Therapy of Children With Autism
3D Object Detection With Latent Support Surfaces
3D Point-Capsule Networks
3D Pose Estimation and 3D Model Retrieval for Objects in the Wild
3D-RCNN: Instance-Level 3D Object Reconstruction via Render-and-Compare
3D Registration of Curves and Surfaces Using Local Differential Information
3D Semantic Segmentation With Submanifold Sparse Convolutional Networks
3D Semantic Trajectory Reconstruction From 3D Pixel Continuum
4DFAB: A Large Scale 4D Database for Facial Expression Analysis and Biometric Applications
4D Human Body Correspondences From Panoramic Depth Maps
A2-RL: Aesthetics Aware Reinforcement Learning for Image Cropping
ABC: A Big CAD Model Dataset For Geometric Deep Learning
A Bi-Directional Message Passing Model for Salient Object Detection
A Biresolution Spectral Framework for Product Quantization
A Causal And-Or Graph Model for Visibility Fluent Reasoning in Tracking Interacting Objects
Accel: A Corrective Fusion Network for Efficient Semantic Segmentation on Video
Accurate and Diverse Sampling of Sequences Based on a "Best of Many" Sample Objective
A Certifiably Globally Optimal Solution to the Non-Minimal Relative Pose Problem
A Closer Look at Spatiotemporal Convolutions for Action Recognition
A Common Framework for Interactive Texture Transfer
A Constrained Deep Neural Network for Ordinal Regression
Action Sets: Weakly Supervised Action Segmentation Without Ordering Constraints
Active Fixation Control to Predict Saccade Sequences
Actor and Action Video Segmentation From a Sentence
Actor and Observer: Joint Modeling of First and Third-Person Videos
AdaDepth: Unsupervised Content Congruent Adaptation for Depth Estimation
A Deeper Look at Power Normalizations
Adversarial Complementary Learning for Weakly Supervised Object Localization
Adversarial Data Programming: Using GANs to Relax the Bottleneck of Curated Labeled Data
Adversarial Feature Augmentation for Unsupervised Domain Adaptation
Adversarially Learned One-Class Classifier for Novelty Detection
Adversarially Occluded Samples for Person Re-Identification
A Face-to-Face Neural Conversation Model
A Fast Resection-Intersection Method for the Known Rotation Problem
A Generative Adversarial Approach for Zero-Shot Learning From Noisy Texts
A Generative Appearance Model for End-to-end Video Object Segmentation
A Hierarchical Generative Model for Eye Image Synthesis and Eye Gaze Estimation
A High-Quality Denoising Dataset for Smartphone Cameras
A Hybrid l1-l0 Layer Decomposition Model for Tone Mapping
Aligning Infinite-Dimensional Covariance Matrices in Reproducing Kernel Hilbert Spaces for Domain Adaptation
Alive Caricature From 2D to 3D
A Low Power, High Throughput, Fully Event-Based Stereo System
Alternating-Stereo VINS: Observability Analysis and Performance Evaluation
A Memory Network Approach for Story-Based Temporal Summarization of 360° Videos
A Minimalist Approach to Type-Agnostic Detection of Quadrics in Point Clouds

Page 1/21 next »