CVPR 2016 Summary Reading

to illustrate the summary of all the articles in CVPR2016, summarize, the summary only retains the innovation point part. ORAL SESSION Image captioning and Question answering Monday, June 27th, 9:00am-10:05am.

These papers'll also be presented at the following poster session

1 Deep compositional captioning:describing novel Object Categories without paired Training Data.

Lisa Anne Hendricks, Subhashini Venugopalan, Marcus Rohrbach, Raymond Mooney, Kate Saenko, Trevor Darrell

The general job is that they can also describe objects that are not present in the picture-statement library. In this work, we propose the deep compositional captioner (DCC) to address the task of generating descriptions of novel OB Jects which is not present in paired image-sentence datasets.

2 Generation and comprehension of unambiguous Object descriptions.

Junhua Mao, Jonathan Huang, Alexander Toshev, Oana Camburu, Alan L. Yuille, Kevin Murphy

is also a form of image description, the author points out that the image description can be an objective evaluation index. We propose a method that can generate an unambiguous description (known as a referring expression) of a specific object or Region in an image, and which can also comprehend or interpret such a expression to infer which object is being describe D.

3 Stacked Attention Networks for Image Question answering.

Zichao Yang, Xiaodong He, Jianfeng Gao, Li Deng, Alex Smola

The article is used for the image question and answer, for example the question picture has several people, makes the corresponding reply, feels more difficult. The innovation here is a stack-based network. This paper presents stacked attention networks (SANs) is learn to answer natural language questions from images.

4 Image Question answering Using convolutional neural Network with Dynamic Parameter prediction.

Hyeonwoo Noh, Paul hongsuck Seo, Bohyung Han

Image question and answer, the innovation here is to add an adaptive parameter layer, which uses GRU learning from the applicable parameters. We tackle Image question answering (IMAGEQA) problem by learning a convolutional neural network (CNN) with a dynamic param Eter layer whose weights is determined adaptively based on questions.

5 Neural Module Networks.

Jacob Andreas, Marcus Rohrbach, Trevor Darrell, Dan Klein

The goal is also the image question and answer, the innovation is simultaneously considers two aspects content: expresses the question and the language model (embarrassed, not all simultaneously considers these two aspects). Not carefully. Our approach decomposes questions into their linguistic substructures, and uses these structures to dynamically instantiat E Modular networks (with reusable components for recognizing dogs, classifying colors, etc.)

Go back to Calendar >

SPOTLIGHT SESSION Language and Vision Monday, June 27th, 10:05am-10:30am.

These papers'll also be presented at the following poster session

6 learning deep representations of fine-grained Visual descriptions.

Scott Reed, Zeynep Akata, Honglak Lee, Bernt Schiele

Deal with zero-shot problem, not carefully understand the innovation point. is basically divided into two parts, from the topic, in fact, the use of deep learning to obtain fine-grained characteristics of the expression. Our proposed models train end-to-end to align with the fine-grained and category-specific content of images. Natural language provides a flexible and compact by the encoding only the salient visual aspects for distinguishing Catego Ries.

7 multi-cue Zero-shot Learning with strong supervision.

Zeynep Akata, Mateusz Malinowski, Mario Fritz, Bernt Schiele

Although it is zero-shot, it is irrelevant. It is not a way to deal with deep learning.

8 latent embeddings for zero-shot classification.

Yongqin Xian, Zeynep Akata, Gaurav Sharma, Quynh Nguyen, Matthias Hein, Bernt Schiele

Although it is zero-shot, it is irrelevant. It is not a way to deal with deep learning.

9 One-Shot learning of Scene Locations via Feature trajectory Transfer.

Roland Kwitt, Sebastian hegenbart, Marc Niethammer

Although it is one-shot, it is irrelevant. It is not a way to deal with deep learning.

Ten learning Attributes Equals multi-source Domain generalization.

Chuang Gan, Tianbao Yang, boqing Gong

It is not a way to deal with deep learning.

anticipating Visual representations from unlabeled Video.

Carl Vondrick, Hamed pirsiavash, Antonio Torralba

The core content is the use of deep learning to predict the next time or time period of behavioral activities. We present a framework that capitalizes on temporal structure in unlabeled video to learn to anticipate human actions and Objects. The key idea behind our approach is so we can train deep networks to predict the visual representation of images in the Future.

Go back to Calendar >

ORAL SESSION Matching and Alignment Monday, June 27th, 9:00am-10:05am.

These papers'll also be presented at the following poster session

Learning to Assign orientations to Feature Points.

Kwang Moo Yi, Yannick Verdie, Pascal Fua, Vincent Lepetit

The content of this paper is to use deep learning to calibrate the feature points, to match the algorithm, and to propose a new activation function. We show how to train a convolutional neural Network to assign a canonical orientation to feature points given an image pat Ch centered on the feature point.

Learning dense correspondence via 3d-guided Cycle consistency.

Tinghui Zhou, Philipp Krähenbuhl, Mathieu Aubry, qixing Huang, Alexei A. Efros

It is also the use of deep learning to explore cross-instance similarity. We exploit this consistency as a supervisory signal to train a convolutional neural network to predict Cross-instance Corr Espondences between pairs of images depicting objects of the same category.

The Global Patch Collider.

Shenlong Wang, Sean Ryan Fanello, Christoph Rhemann, Shahram Izadi, Pushmeet Kohli

Not deep learning.

Joint Probabilistic Matching Using m-best Solutions.

Seyed Hamid Rezatofighi, Anton Milan, Zhen Zhang, Qinfeng Shi, Anthony Dick, Ian Reid

Not deep learning.

Alignment Across Large poses:a 3D solution.

Xiangyu Zhu, Zhen Lei, xiaoming Liu, Hailin Shi, Stan Z. Li

This paper presents a three-dimensional face correction technique using deep learning. We propose a solution to the three problems on an new alignment framework, called 3D dense face Alignment (3DDFA), in whic H a dense 3D face model was fitted to the image via convolutional neutral network (CNN)

Go back to Calendar >

SPOTLIGHT SESSION Segmentation and Contour Detection Monday, June 27th, 10:05am-10:30am.

These papers'll also be presented at the following poster session

Interactive segmentation on RGBD Images via Cue Selection.

Jie Feng, Brian Price, Scott Cohen, Shih-fu Chang

Not deep learning.

layered Scene decomposition via the OCCLUSION-CRF.

Chen Liu, Pushmeet Kohli, Yasutaka Furukawa

Not deep learning.

Affinity cnn:learning pixel-centric pairwise relations for Figure/ground embedding.

Michael Maire, Takuya Narihira, Stella x. Yu

Get a affinity matrix through deep learning. Interest is not enough. We train a convolutional neural network (CNN) to directly predict the pairwise relationships it define this affinity mat Rix

weakly supervised Object boundaries.

Anna Khoreva, Rodrigo Benenson, Mohamed Omran, Matthias Hein, Bernt Schiele

Not deep learning.

Object Contour Detection with a Fully convolutional encoder-decoder Network.

Jimei Yang, Brian price, Scott Cohen, Honglak Lee, Ming-hsuan Yang

A contour recognition algorithm based on full convolutional coding and decoding network is proposed. We develop a deep learning algorithm for contour detection with a fully convolutional encoder-decoder network.

Go back to Calendar >

POSTER SESSION Poster Session 1-1. Monday, June 27th, 10:30am-12:30pm. Images and Language

What Value does Explicit high level concepts has in the Vision to Language problems?.

Qi Wu, Chunhua Shen, Lingqiao Liu, Anthony Dick, Anton van den Hengel

The article considers the image description, but because the existing methods are directly mapping objects in the image into textual information, and do not consider the high-level semantic information, the article puts forward the innovation point of considering the high-level information. We propose a method of incorporating high-level concepts into the successful CNN-RNN approach, and show that it achieves a  Significant improvement on the State-of-the-art in both image captioning and visual question answering. Edge Contour Detection

Detection of curved Edges at low SNR.

Nati Ofir, Meirav Galun, Boaz Nadler, Ronen Basri

Not deep learning. Solving the problem is equivalent to the edge detection while eliminating noise.

The Object Skeleton Extraction in Natural Images is fusing scale-associated deep Side Outputs.

Wei Shen, Kai Zhao, Yuan Jiang, Yan Wang, Zhijiang Zhang, Xiang Bai

A skeleton extraction algorithm is proposed using the full convolutional network, which seems to be the article in the previous blog. In this paper, we present a fully convolutional network with multiple scale-associated side outputs to address this proble M. by observing the relationship between the receptive field sizes of the sequential stages in the network and the Skeleto N Scales They can capture, we introduce a scale-associated side output to each stage

Learning Relaxed Deep supervision for Better Edge Detection.

Yu Liu, Michael S. Lew

An edge detection algorithm based on deep learning. We propose using relaxed deep supervision (RDS) within convolutional neural networks for edge detection.

occlusion boundary Detection via Deep Exploration of the Context.

Huan Fu, Chaohui Wang, Dacheng Tao, Michael J. Black

Occlusion Edge detection. Based on deep learning in this paper, we improve occlusion boundary detection via enhanced exploration of contextual information (e.g., lo Cal Structural boundary patterns, observations from surrounding regions, and temporal context), and in doing so develop a Novel approach based on convolutional neural Networks (CNNs) and conditional random fields (CRFs)

semicontour:a semi-supervised Learning approach for Contour Detection.

Zizhao Zhang, Fuyong Xing, Xiaoshuang Shi, Lin Yang

Not deep learning. Contour detection. Feature Extraction and Description

Learning to Localize Little landmarks.

Saurabh Singh, Derek hoiem, David Forsyth

Not deep learning, feature extraction.

interactive:inter-layer activeness propagation.

Lingxi Xie, Liang Zheng, Jingdong Wang, Alan L. Yuille, Qi Tian

Equivalent to an improvement of the activation function. In this paper, we present InterActive, a novel algorithm which computes the activeness of neurons and network connections.

Exploit bounding Box Annotations for multi-label Object recognition.

Hao Yang, Joey Tianyi Zhou, Yu Zhang, Bin-bin Gao, Jianxin Wu, Jianfei Cai

A multi-target recognition algorithm is used to identify deep learning in multi-label target recognition. In this paper, we incorporate local information to enhance the feature discriminative power.

ti-pooling:transformation-invariant POOLING for Feature learning in convolutional neural Networks.

Dmitry Laptev, Nikolay Savinov, Joachim M. Buhmann, Marc Pollefeys

A new pooling operator. In this paper we present a deep neural network topology, incorporates a simple to implement Transformation-invariant p Ooling operator (ti-pooling)

Fashion Style in Floats:joint Ranking and classification Using Weak Data for Feature Extraction.

Edgar Simo-serra, Hiroshi Ishikawa

Not deep learning.

equiangular Kernel Dictionary Learning with applications to Dynamic Texture analysis.

Yuhui Quan, Chenglong Bao, Hui Ji

Not deep learning.

The Compact Bilinear Pooling.

Yang Gao, Oscar Beijbom, Ning Zhang, Trevor Darrell

Not deep learning. Feature Extraction and Matching

accumulated stability voting:a Robust descriptor from descriptors of multiple Scales.

Tsun-yi Yang, Yen-yu Lin, Yung-yu Chuang

Not deep learning.

Comal:good Features to Match on Object boundaries.

Swarna K. Ravindran, Anurag Mittal

Not deep learning.

PNS Progressive Feature Matching with Alternate descriptor Selection and correspondence enrichment.

Yuan-ting Hu, Yen-yu Lin

Not deep learning. Image Segmentation

A New Finsler Minimal Path Model with curvature penalization for Image segmentation and Closed Contour Detectio N.

Da Chen, Jean-marie Mirebeau, Laurent d. Cohen

Not deep learning.

scale-aware Alignment of hierarchical Image segmentation.

Yuhua Chen, Dengxin Dai, Jordi pont-tuset, Luc Van Gool

Not deep learning.

The deep Interactive Object Selection.

Ning Xu, Brian Price, Scott Cohen, Jimei Yang, Thomas S. Huang

Target recognition for deep learning interactions. Choose. In this paper, we present a novel deep-learning-based algorithm which have much better understanding of objectness and can Reduce user interactions to just a few clicks.

The Plug?   Predicting If Computers or humans should Segment Images.

Danna Gurari, Suyog Jain, Margrit Betke, Kristen Grauman

Not deep learning.

the Shadows, Shape Priors shine:using occlusion to Improve multi-region segmentation.

Yuka Kihara, Matvey Soloviev, Tsuhan Chen

Deep learning, but not much interest.

convexity Shape Constraints for Image segmentation.

Loic A. Royer, David L. Richmond, Carsten Rother, Bjoern Andres, Dagmar Kainmueller

Not deep learning.

MCMC Shape sampling for Image segmentation with nonparametric shape Priors.

Ertunc Erdil, Sinan Yildirim, Müjdat Cetin, Tolga Tasdizen

Not deep learning. Low-level Vision

Noise Modeling to Blind Image denoising.

Fengyuan Zhu, Guangyong Chen, Pheng-ann Heng

Not deep learning.

efficient and robust Color consistency for Community Photo collections.

Jaesik Park, yu-wing Tai, Sudipta N. Sinha, in so Kweon

Not deep learning.

needle-match:reliable Patch Matching under high uncertainty.

Or Lotan, Michal Irani

Not deep learning.

reconnet:non-iterative reconstruction of Images from compressively sensed measurements.

Kuldeep Kulkarni, Suhas lohit, Pavan Turaga, Ronan Kerviche, Amit Ashok

The recovery and reconstruction of compressed-sensing image is made by convolution neuron network. We propose a novel convolutional neural Network (CNN) architecture which takes in CS measurements of an image as input and Outputs an intermediate reconstruction

soft-segmentation Guided Object Motion Deblurri