DenseNet for dense flow

By Yi Zhu, Shawn D. Newsam. International Conference on Information Photonics, 2017

305 citations recorded by the source corpus.

Classical approaches for estimating optical flow have achieved rapid progress in the last decade. However, most of them are too slow to be applied in real-time video analysis. Due to the great success of deep learning, recent work has focused on using CNNs to solve such dense prediction problems. In this paper, we investigate a new deep architecture, Densely Connected Convolutional Networks (DenseNet), to learn optical flow. This specific architecture is ideal for the problem at hand as it provides shortcut connections throughout the network, which leads to implicit deep supervision. We extend current DenseNet to a fully convolutional network to learn motion estimation in an unsupervised manner. Evaluation results on three standard benchmarks demonstrate that DenseNet is a better fit than other widely adopted CNN architectures for optical flow estimation.

Subfields: Computer Vision and Pattern Recognition, Radiology, Nuclear Medicine and Imaging

View Paper: https://www.semanticscholar.org/paper/998922bd412d2fcd9a118fdefd32f9989aef9487

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

1. Vishal M. Patel
2. Dacheng Tao
3. Ming-Hsuan Yang
4. Luc Van Gool
5. Yi Yang
6. Trevor Darrell
7. Andrew Zisserman
8. Jitendra Malik

In Radiology, Nuclear Medicine and Imaging

1. He Zhang
2. Vishal M. Patel
3. Stefan Roth
4. Junhwa Hur
5. Frank Hutter
6. Thomas Brox
7. Junaid Qadir
8. Adeel Razi

Citations in the corpus

• Iterative Residual Refinement for Joint Optical Flow and Occlusion Estimation (2019, background): “To alleviate the need for training data with ground truth in a specific domain, unsupervised [1, 40, 47, 58, 63, 66] and semisupervised [31, 65] alternatives have also been developed.”
• UnFlow: Unsupervised Learning of Optical Flow with a Bidirectional Census Loss (2017, background): “Other subsequent work (Zhu and Newsam 2017) attempted to improve unsupervised CNNs by replacing the underlying FlowNet architecture with a different network.”
• Leveraging Data Science to Combat COVID-19: A Comprehensive Review (2020, methodology): “[143] Chest X-rays DenseNet Used DenseNet Architecture [144] for COVID-19 detection in X-rays.”
• AutoDispNet: Improving Disparity Estimation With AutoML (2019, background): “Several hand-designed encoderdecoder architectures have emerged [2, 28, 77, 80] which incorporate the above architecture design choices.”
• Densely Connected Pyramid Dehazing Network (2018, methodology): “To reconstruct the transmission map into the original resolution, we stack five dense blocks with the refined up-sampling Transition-Blocks [19, 59] as the decoding module.”
• Hidden Two-Stream Convolutional Networks for Action Recognition (2017, background): “earn such a motion estimation model: { We need to train the models without supervision. The ground truth ow required for supervised training is usually not available except for limited synthetic data [34,36,37]. { We need to train our optical ow estimation models from scratch. The models (lters) lea”