unizard
commented
6 years ago Computer Vision Papers in NIPS2017
The number of vision papers : 42
Image Retrieval (#paper: 1) Recognition (#paper: 14) Data and Label (#paper: 4) Novel Architecture (#paper: 9) Low-level Vision (#paper: 10) ETC (#paper: 4)
@ Image Retrieval (#paper: 2)
Fast Search
@ Recognition (#paper: 14)
VQA
- Best of Both Worlds: Transferring Knowledge from Discriminative Learning to a Generative Visual Dialog Model, GIT, FAIR
- Multimodal Learning and Reasoning for Visual Question Answering, NUS
- Visual Reference Resolution using Attention Memory for Visual Dialog, Postech, Disney Research
- High-Order Attention Models for Visual Question Answering
Captioning
- Contrastive Learning for Image Captioning, CUHK
Visual Genome - Relationship Estimation
- Pixels to Graphs by Associative Embedding, UMICH.
- dataset: http://visualgenome.org/
- Stanford, Fei-Fei Li
ActionRecognition
- Pose Guided Person Image Generation, MPI, ETH, Luc Van Gool
- Self-supervised Learning of Motion Capture, CMU, Adobe
- PredRNN: Recurrent Neural Networks for Predictive Learning using Spatiotemporal LSTMs, Tsinghua
- Associative Embedding: End-to-End Learning for Joint Detection and Grouping, Pose-AE Code, UMICH.
Text Recognition/OCR
- Gated Recurrent Convolution Neural Network for OCR, [PDF], Tsinghua
3D Recognition
- PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space, Stanford.
3D Modeling
- Learning a Multi-View Stereo Machine, UC Berkely, Jitendra Malik
- MarrNet: 3D Shape Reconstruction via 2.5D Sketches, MIT, Bill Freeman, Josh Tenenbaum, Projectpage
@ Data and Label (#paper: 4)
Dealing with NoisyData
- Decoupling "when to update" from "how to update"
Dealing with Unlabeled Data
- Max-Margin Invariant Features from Transformed Unlabelled Data
Usage of synthetic data for classification
- Introspective Classification with Convolutional Nets, USCD.
Label Distribution Learning
- Label Distribution Learning Forests, LDLForest, Python, ProjectPage
- About LDL: http://cse.seu.edu.cn/people/xgeng/LDL/index.htm
@ Novel Architecture (#paper: 8)
Relation Network
- A simple neural network module for relational reasoning, Deepmind. (VQA). [***]
- Learning Multiple Tasks with Multilinear Relationship Networks, python, GIT, 제출기록 NIPS2016, ICCV2017, NIPS2017. (Classification)
Knowledge Distillation Network(Teacher=>Student)
- Learning Efficient Object Detection Models with Knowledge Distillation
- Mean teachers are better role models: Weight-averaged consistency targets improve semi-supervised deep learning results, ProjectPage [***Fun]
Catestrophic Forgetting
- Continual Learning with Deep Generative Replay, SKT, MIT (class Incremental Learning)
- Learning multiple visual domains with residual adapters, Oxford, Andrea Vedaldi. (for Multi-task Learning)
Model Compression, Binary CNN
- Towards Accurate Binary Convolutional Neural Network,[PDF]
How to handle rejecting data
- Selective Classification for Deep Neural Networks, *mining for classification
Vector Compression
- Soft-to-Hard Vector Quantization for End-to-End Learning Compressible Representations, ETH, Luc Van Gool
@ Low-level Vision (#paper: 10)
Segmentation
- MaskRNN: Instance Level Video Object Segmentation, [PDF], UIUC ,VT.
- Learning Affinity via Spatial Propagation Networks, Video, NVidia Research, UC Merced. Ming-Hsuan Yang
Optical-Flow/Motion Estimation
- Semi-Supervised Learning for Optical Flow with Generative Adversarial Networks, Optical Flow/Motion, VT.
Intrisinc Image Decomposition
- Self-Supervised Intrinsic Image Decomposition, MIT
Saliency Map
- Real Time Image Saliency for Black Box Classifiers
Image Restoration
- Deep Mean-Shift Priors for Image Restoration, UMD, ProjectPage, Matlab
Image Inpainting
- Learning to Inpaint for Image Compression
Correspondence Search (Depth)
- Matching neural paths: transfer from recognition to correspondence search, Poster, Videos, ETH, M. Pollefeys.
Contour Prediction
- Learning Deep Structured Multi-Scale Features using Attention-Gated CRFs for Contour Prediction
Feature for 360' Camera
- Learning Spherical Convolution for Fast Features from 360° Imagery, UTA, Kristen Grauman
Proceedings https://papers.nips.cc/book/advances-in-neural-information-processing-systems-30-2017 https://github.com/catpanda/NIPS_2017
PaperLists (#Papers 679) https://www.dropbox.com/s/gho79gtg8eo1pez/NIPS2017_PaprList.xlsx?dl=0