[Paper note] FlowNet: Learning Optical Flow with Convolutional Networks
2017-01-11 13:50
387 查看
paper
code
Introduce novel correlation layer
Refine network by upsampling
FlowNetSimple: concatenate two consecutive images.
FlowNetCorr: use correlation layer
Correlation layer
Calculated between two feature maps
c(x1,x2)=∑o∈[−k,k]×[−k,k]<f1(x1+o),f2(x2+o)>
See model picture for an illustration
Refinement
Concatenate the upsampled flow prediction and conv feature map
Middlebury
KITTI
Sintel
Flying Chairs (proposed, auto generated)
Loss function: endpoint error – Euclidean distance between the predicted flow vector and GT.
Conclusion
FlowNet performs a little worse than other OF algorithm, but obviously faster.
Network trained on Flying Chairs (auto generated) data has good generalization ability on natural scenes.
code
Highlight
First paper to use trained CNN for optical flow estimationIntroduce novel correlation layer
Refine network by upsampling
Model
FlowNetSimple: concatenate two consecutive images.
FlowNetCorr: use correlation layer
Correlation layer
Calculated between two feature maps
c(x1,x2)=∑o∈[−k,k]×[−k,k]<f1(x1+o),f2(x2+o)>
See model picture for an illustration
Refinement
Concatenate the upsampled flow prediction and conv feature map
Experiment
Datasets:Middlebury
KITTI
Sintel
Flying Chairs (proposed, auto generated)
Loss function: endpoint error – Euclidean distance between the predicted flow vector and GT.
Conclusion
FlowNet performs a little worse than other OF algorithm, but obviously faster.
Network trained on Flying Chairs (auto generated) data has good generalization ability on natural scenes.
相关文章推荐
- FlowNet: Learning Optical Flow with Convolutional Networks
- CNN光流计算--FlowNet: Learning Optical Flow with Convolutional Networks
- 解读: FlowNet learning optical flow with convolutional networks
- 【论文学习】神经光流网络——用卷积网络实现光流预测(FlowNet: Learning Optical Flow with Convolutional Networks)
- 神经光流网络——用卷积网络实现光流预测(FlowNet: Learning Optical Flow with Convolutional Networks)
- 几分钟走进神奇的光流|FlowNet 2.0: Evolution of Optical Flow Estimation with Deep Networks
- CNN光流计算2--FlowNet 2.0: Evolution of Optical Flow Estimation with Deep Networks
- [Paper note] Language Modeling with Gated Convolutional Networks
- FlowNet 2.0: Evolution of Optical Flow Estimation with Deep Networks
- 解读:FlowNet 2.0 evolution of optical flow estimation with deep networks
- ImageNet Classification with Deep Convolutional Neural Networks翻译总结
- 论文笔记《ImageNet Classification with Deep Convolutional Neural Networks》
- ImageNet Classification with Deep Convolutional Neural Networks
- Learning hierarchical representations for face verification with convolutional deep belief networks
- READING NOTE: Object Detection from Video Tubelets with Convolutional Neural Networks
- Unsupervised Rrpresentation Learning with Deep Convolutional Generative Adversarial Networks
- 论文笔记之:UNSUPERVISED REPRESENTATION LEARNING WITH DEEP CONVOLUTIONAL GENERATIVE ADVERSARIAL NETWORKS
- ImageNet Classification with Deep Convolutional Neural Networks
- [notes] ImageNet Classification with Deep Convolutional Neual NetworkPaper: ImageNet Classification
- ImageNet Classification with Deep Convolutional Neural Networks泛读总结