Oral presentations

The goal of the oral presentations is to carry out a bibliographic study and present the result to the class. You will team in up to two in this work.

Below is a list of papers organized in categories and sub-categories, which can help in finding papers related to each other. Choose a number of papers (not less than two, preferably not more than five) that are related to each other in a clear way. For instance, one is influenced by another, their methods are similar in some respect, they are using the same idea in different ways, or they attack the same problem from different sides. Papers should not be too hard to follow, but they should not be too easy either. They should be interesting.

The list is there only to help in initiating the study, but is by no means a constraint. You are free to start e.g. at arXiv.org CVPR list, CVF CVPR/ICCV open access, or anywhere else. You are also expected to find papers through citations, see below.

Report your choice by Tuesday, December 19. Study the papers in depth. Find how they are connected. Pay attention to related work and try to identify other papers through their citations, that are more connected or more interesting. Feel free at this point to change your list of papers by removing some you found in the list and adding others that you found through studying. But you should not change the subject of your study entirely.

The presentations are on Monday, January 29. You will have 8 minutes each for your talk plus 4 minutes for questions from the class, that is 20 minutes per team in total. You are expected to ask questions at other students' presentations. You will be evaluated based on your choice of papers, the way you have connected them into a story, and how you present them. You should focus on the main ideas. Implementation details are irrelevant for instance, but your talk shouldn't be shallow either. Your goal is to present something interesting to your fellow students, so they learn something from your talk.

Action recognition
Recurrent neural networks
RNN Fisher Vectors for Action Recognition and Image Annotation
Beyond Temporal Pooling: Recurrence and Temporal Convolutions for Gesture Recognition in Video
Chained Predictions Using Convolutional Neural Networks
Beyond Short Snippets: Deep Networks for Video Classification
Long-Term Recurrent Convolutional Networks for Visual Recognition and Description
Unsupervised Learning of Video Representations Using LSTMs
Modeling Spatial-Temporal Clues in a Hybrid Deep Learning Framework for Video Classification
Temporal convolution
Beyond Temporal Pooling: Recurrence and Temporal Convolutions for Gesture Recognition in Video
ConvNet Architecture Search for Spatiotemporal Feature Learning
Learning Spatiotemporal Features with 3D Convolutional Networks
Convolutional Two-Stream Network Fusion for Video Action Recognition
Pooling
Neural Aggregation Network for Video Face Recognition
Actionvlad: Learning Spatio-Temporal Aggregation for Action Classification
Deep Temporal Linear Encoding Networks
Higher-Order Pooling of CNN Features via Kernel Linearization for Action Recognition
Second-Order Temporal Pooling for Action Recognition
Generalized Rank Pooling for Activity Recognition
Temporal Pyramid Pooling Based Convolutional Neural Networks for Action Recognition
Order-Aware Convolutional Pooling for Video Based Action Recognition
Classification
Fine-grained
Object-Centric Sampling for Fine-Grained Image Classification
Bilinear CNN Models for Fine-Grained Visual Recognition
Mask-CNN: Localizing Parts and Selecting Descriptors for Fine-Grained Image Recognition
Deep Deformation Network for Object Landmark Localization
Fine-grained Recognition in the Wild: A Multi-Task Domain Adaptation Approach
Mining Discriminative Triplets of Patches for Fine-Grained Classification
Fine-Grained Pose Prediction, Normalization, and Recognition
Part Detector Discovery in Deep Convolutional Neural Networks
Part-Stacked CNN for Fine-Grained Visual Categorization
Convolution
On graphs
Unifying Local and Non-Local Signal Processing with Graph CNNs
Graph Based Convolutional Neural Network
Dynamic Filters in Graph Convolutional Networks
Dynamic Edge-Conditioned Filters in Convolutional Neural Networks on Graphs
A Generalization of Convolutional Neural Networks to Graph-Structured Data
Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering
Spectral Networks and Locally Connected Networks on Graphs
Diffusion-Convolutional Neural Networks
Deep Convolutional Networks on Graph-Structured Data
Adaptive
Deformable Convolutional Networks
3D Morphable Models as Spatial Transformer Networks
Spatial Transformer Networks
Active Convolution: Learning the Shape of Convolution for Image Classification
Learning Dilation Factors for Semantic Segmentation of Street Scenes
Dynamic Filter Networks
Dense Transformer Networks
Dynamic Steerable Blocks in Deep Residual Networks
Irregular Convolutional Neural Networks
Sparse
Training Sparse Neural Networks
PerforatedCNNs: Acceleration Through Elimination of Redundant Convolutions
Sparsity Invariant CNNs
Generalizing the Convolution Operator to Extend CNNs to Irregular Domains
Submanifold Sparse Convolutional Networks
Why Regularized Auto-Encoders Learn Sparse Representation;
Superpixel Convolutional Networks Using Bilateral Inceptions
Sparse Convolutional Networks Using the Permutohedral Lattice
Quantization
XNOR-Net: Imagenet Classification Using Binary Convolutional Neural Networks
Binaryconnect: Training Deep Neural Networks with Binary Weights During Propagations
Ternary Weight Networks
Local Binary Convolutional Neural Networks
Incremental Network Quantization: Towards Lossless CNNs with Low-Precision Weights
Deep Neural Networks Are Robust to Weight Binarization and Other Non-Linear Distortions
Compressing Deep Convolutional Networks Using Vector Quantization
Deep Compression: Compressing Deep Neural Networks with Pruning, Trained Quantization and Huffman Coding
Dilated
Multi-Scale Context Aggregation By Dilated Convolutions
Rethinking Atrous Convolution for Semantic Image Segmentation
Dilated Residual Networks
Dense CNN Learning with Equivalent Mappings
In frequency
Fast Neural Networks with Circulant Projections
Shapenet: Convolutional Neural Networks on Non-Euclidean Manifolds
Fast Training of Convolutional Networks Through FFTs
Fast Convolutional Nets with Fbfft: A GPU Performance Evaluation
Structured
Scaling the Scattering Transform: Deep Hybrid Networks
Structured Convolution Matrices for Energy-Efficient Deep Learning
Structured Receptive Fields in CNNs
Gabor Convolutional Networks
Low-rank
Training CNNs with Low-Rank Filters for Efficient Image Classification
Exploiting Local Structures with the Kronecker Layer in Convolutional Networks
Convolutional Neural Networks with Low-Rank Regularization
Interleaved Group Convolutions for Deep Neural Networks
Coordinating Filters for Faster Deep Neural Networks
Speeding-Up Convolutional Neural Networks Using Fine-Tuned CP-Decomposition
Factorized Convolutional Neural Networks
Deep Learning with Separable Convolutions
Decomposeme: Simplifying Convnets for End-to-End Learning
Detection
Counting
Learning to Point and Count
Deepsetnet: Predicting Sets with Deep Neural Networks
Crowd counting via scale-adaptive convolutional neural network
Object proposals
Boosting Convolutional Features for Robust Object Proposals
Scale-Aware Pixel-Wise Object Proposal Networks
Scalable Object Detection Using Deep Neural Networks
Object-Proposal Evaluation Protocol Is Gameable
ME R-CNN: Multi-Expert Region-Based CNN for Object Detection
Learning to Segment Object Candidates
Deepproposals: Hunting Objects and Actions By Cascading Deep Convolutional Layers
Active Object Localization with Deep Reinforcement Learning
Deepbox: Learning Objectness with Convolutional Networks
DeepProposal: Hunting Objects By Cascading Deep Convolutional Layers
Part-based
Deepm: A Deep Part-Based Model for Object Detection and Semantic Part Localization
Deepid-Net: Deformable Deep Convolutional Neural Networks for Object Detection
Mask-CNN: Localizing Parts and Selecting Descriptors for Fine-Grained Image Recognition
Do Semantic Parts Emerge in Convolutional Neural Networks;
Actions and Attributes from Wholes and Parts
Deformable Part Models Are Convolutional Neural Networks
Object Detection Networks on Convolutional Feature Maps
Neural Activation Constellations: Unsupervised Part Model Discovery with Convolutional Networks
Joint Object and Part Segmentation Using Deep Learned Potentials
Deformable Part-based Fully Convolutional Network for Object Detection
Part-Stacked CNN for Fine-Grained Visual Categorization
Non-maxima suppresion
Improving Object Detection with One Line of Code
A Convnet for Non-Maximum Suppression
Learning Detection with Diverse Proposals
Learning Non-Maximum Suppression
Pyramid architecture
Feature Pyramid Networks for Object Detection
Beyond Skip Connections: Top-Down Modulation for Object Detection
Focal Loss for Dense Object Detection
DSSD : Deconvolutional Single Shot Detector
Attention-guided
Attend Refine Repeat: Active Box Proposal Generation via in-Out Localization
Pronet: Learning to Propose Object-Specific Boxes for Cascaded Neural Networks
Hierarchical Object Detection with Deep Reinforcement Learning
Multiple Object Recognition with Visual Attention
Attentional Network for Visual Object Detection
G-CNN: an Iterative Grid Based Object Detector
Learning to Decompose for Object Detection and Instance Segmentation
Spatial Memory for Context Reasoning in Object Detection
Adaptive Object Detection Using Adjacency and Zoom Prediction
Weakly supervised
Weakly Supervised Deep Detection Networks
Multiple Instance Detection Network with Online Instance Classifier Refinement
Improving Weakly-Supervised Object Localization By Micro-Annotation
Learning Semantic Part-Based Models From Google Images
Constrained Convolutional Neural Networks for Weakly Supervised Segmentation
Contextlocnet: Context-Aware Deep Network Models for Weakly Supervised Localization
We Don t Need No Bounding-Boxes: Training Object Class Detectors Using Only Human Verification
From Image-Level to Pixel-Level Labeling with Convolutional Networks
Neural Activation Constellations: Unsupervised Part Model Discovery with Convolutional Networks
Self Paced Deep Learning for Weakly Supervised Object Detection
Proposal-based
Training Region-Based Object Detectors with Online Hard Example Mining
CoupleNet: Coupling Global Structure with Local Parts for Object Detection
R-FCN: Object Detection via Region-Based Fully Convolutional Networks
Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation
PVANET: Deep But Lightweight Neural Networks for Real-Time Object Detection
Hypernet: Towards Accurate Region Proposal Generation and Joint Object Detection
Denet: Scalable Real-Time Object Detection with Directed Sparse Sampling
Fast R-CNN
Inside-Outside Net: Detecting Objects in Context with Skip Pooling and Recurrent Neural Networks
Face
Recognition
Deeply Learned Face Representations Are Sparse, Selective, and Robust
Face Identification with Bilinear CNNs
Megaface: A Million Faces for Recognition at Scale
Multi-View Face Detection Using Deep Convolutional Neural Networks
When Face Recognition Meets with Deep Learning: an Evaluation of Convolutional Neural Networks for Face Recognition
Towards the Design of an End-to-End Automated System for Image and Video-Based Recognition
Beyond Frontal Faces: Improving Person Recognition Using Multiple Cues
Deep Learning Face Representation By Joint Identification-Verification
Invariance
Rotation
Flip-Rotate-Pooling Convolution and Split Dropout on Convolution Neural Networks for Image Classification
Rotation Equivariant Vector Field Networks
Scaling the Scattering Transform: Deep Hybrid Networks
Rotation-Invariant Convolutional Neural Networks for Galaxy Morphology Prediction
Harmonic Networks: Deep Translation and Rotation Equivariance
Polar Transformer Networks
Deep Rotation Equivariant Network
Exploiting Cyclic Symmetry in Convolutional Neural Networks
TI-POOLING: Transformation-Invariant Pooling for Feature Learning in Convolutional Neural Networks
Gabor Convolutional Networks
Scale
Learning Scale-Variant and Scale-Invariant Features for Deep Image Classification
Fast Training of Convolutional Neural Networks via Kernel Rescaling
Competitive Multi-Scale Convolution
Scale-Invariant Convolutional Neural Network
Looking Outside of the Box: Object Detection and Localization with Multi-Scale Patterns
Autoscaler: Scale-Attention Networks for Visual Correspondence
What Happened to My Dog in That Network: Unraveling Top-Down Generators in Convolutional Neural Networks
Neural Multigrid
Attention to Scale: Scale-Aware Semantic Image Segmentation
Locally Scale-Invariant Convolutional Neural Networks
Learning Across Scales : A Multiscale Method for Convolution Neural Networks
Measuring
What Can We Learn About CNNs From a Large Scale Controlled Object Dataset;
Manitest: Are Classifiers Really Invariant;
How Well Can a CNN Marginalize Simple Nuisances It Is Designed for;
Generalization Error of Invariant Classifiers
Exploring Invariances in Deep Convolutional Neural Networks Using Synthetic Images
Geodesics of Learned Representations
Mirror, Mirror on the Wall, Tell Me, Is the Error Small;
What Is Holding Back Convnets for Detection;
Registration/correspondence
Semantic parts
Universal Correspondence Network
FCSS: Fully Convolutional Self-Similarity for Dense Semantic Correspondence
Scnet: Learning Semantic Correspondence
Anchornet: A Weakly Supervised Network to Learn Geometry-Sensitive Features for Semantic Matching
Dense optical flow
Optical Flow Estimation Using a Spatial Pyramid Network
Patchbatch: a Batch Augmented Loss for Optical Flow
Unsupervised Feature Learning for Dense Correspondences Across Scenes
Learning Image Matching By Simply Watching Video
Autoscaler: Scale-Attention Networks for Visual Correspondence
Deep Convolutional Matching
Learning to Extract Motion From Videos in Convolutional Neural Networks
Fully-Trainable Deep Matching
Deep Semantic Matching for Optical Flow
Flownet: Learning Optical Flow with Convolutional Networks
Flownet 2.0: Evolution of Optical Flow Estimation with Deep Networks
Retrieval
Spatial matching
Deep Region Hashing for Efficient Large-Scale Instance Search From Images
Particular Object Retrieval with Integral Max-Pooling of CNN Activations
CNN Features Off-the-Shelf: an Astounding Baseline for Recognition
Visual Instance Retrieval with Deep Convolutional Networks
CNN Features Off-the-Shelf: an Astounding Baseline for Recognition
Image Retrieval with Deep Local Features and Attention-Based Keypoints
Voronoi-Based Compact Image Descriptors: Efficient Region-of-Interest Retrieval with VLAD and Deep-Learning-Based Descriptors
Pooling
Object Level Deep Feature Pooling for Compact Image Representation
Embedding Based on Function Approximation for Large Scale Image Search
Aggregating Deep Convolutional Features for Image Retrieval
Context Aware Query Image Representation for Particular Object Retrieval
Deep Convolutional Features for Image Based Retrieval and Scene Categorization
Image Retrieval with Deep Local Features and Attention-Based Keypoints
Netvlad: CNN Architecture for Weakly Supervised Place Recognition
Faster R-CNN Features for Instance Search
Where to Focus: Query Adaptive Matching for Instance Retrieval Using Convolutional Feature Maps
Learning similarity
Learning Non-Metric Visual Similarity for Image Retrieval
CNN Image Retrieval Learns From Bow: Unsupervised Fine-Tuning with Hard Examples
Fine-Tuning CNN Image Retrieval with No Human Annotation
Deep Image Retrieval: Learning Global Representations for Image Search
End-to-End Learning of Deep Visual Representations for Image Retrieval
Segmentation
Weakly supervised
Weakly Supervised Semantic Labelling and Instance Segmentation
Weakly-Supervised Semantic Segmentation Using Motion Cues
Exploiting Saliency for Object Segmentation From Image Level Labels
What s the Point: Semantic Segmentation with Point Supervision
Edge-based
Convolutional Oriented Boundaries
Semantic Image Segmentation with Task-Specific Edge Detection Using CNNs and a Discriminatively Trained Domain Transform
One-Shot Video Object Segmentation
Convolutional Oriented Boundaries: From Image Segmentation to High-Level Tasks
Instance segmentation
Shape-Aware Instance Segmentation
Instance-Sensitive Fully Convolutional Networks
Object Boundary Guided Semantic Segmentation
Fully Convolutional Instance-Aware Semantic Segmentation
Object Detection Free Instance Segmentation with Labeling Transformations
Learning to Segment Object Candidates
A Multipath Network for Object Detection
Monocular Object Instance Segmentation and Depth Ordering with CNNs
Mask R-CNN
Semantic Instance Segmentation via Deep Metric Learning
Pixelwise Instance Segmentation with a Dynamically Instantiated Network
Deconvolution
Multi-Scale Context Aggregation By Dilated Convolutions
Large Kernel Matters -: Improve Semantic Segmentation By Global Convolutional Network
The One Hundred Layers Tiramisu: Fully Convolutional Densenets for Semantic Segmentation
Recombinator Networks: Learning Coarse-to-Fine Feature Aggregation
Global Deconvolutional Networks for Semantic Segmentation
Pixel Deconvolutional Networks
Fully Convolutional Networks for Semantic Segmentation
Decoupled Deep Neural Network for Semi-Supervised Semantic Segmentation
Learning Deconvolution Network for Semantic Segmentation
Deconvolutional Feature Stacking for Weakly-Supervised Semantic Segmentation
Video
Deep End2End Voxel2Voxel Prediction
Clockwork Convnets for Video Semantic Segmentation
Unsupervised Object Segmentation in Video By Efficient Selection of Highly Probable Positive Features
Learning Video Object Segmentation with Visual Memory
Spatio-Temporal Image Boundary Extrapolation
Recurrent Fully Convolutional Networks for Video Segmentation
Object Detection, Tracking, and Motion Segmentation for Object-Level Video Segmentation
Morphology
Flood-Filling Networks
Deep Watershed Transform for Instance Segmentation
Reformulating Level Sets As Deep Recurrent Neural Network Approach to Semantic Segmentation
Learned Watershed: End-to-End Learning of Seeded Segmentation
Conditional random fields
Semantic Image Segmentation with Deep Convolutional Nets and Fully Connected CRFs
Discriminative Training of Deep Fully-Connected Continuous CRF with Task-Specific Loss
Fast Semantic Image Segmentation with High Order Context and Guided Filtering
Deepcut: Object Segmentation From Bounding Box Annotations Using Convolutional Neural Networks
Learning Transferrable Knowledge for Semantic Segmentation with Deep Convolutional Neural Network
Deeplab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs
Weakly- and Semi-Supervised Learning of a DCNN for Semantic Image Segmentation
Conditional Random Fields As Recurrent Neural Networks
Efficient Piecewise Training of Deep Structured Models for Semantic Segmentation
Things and stuff
Convolutional Feature Masking for Joint Object and Stuff Segmentation
Weakly Supervised Object Localization Using Things and Stuff Transfer
COCO-Stuff: Thing and Stuff Classes in Context
Stuffnet: Using Stuff to Improve Object Detection