1
3D Vision, Spring Semester 2016 Goa l: Requirements / Tools: Supervisor: Descriptio n: Goal: Description: 2018 Transfer from Recognition to Optical Flow by Matching Neural Paths Implementation of Optical Flow Method by Matching Neural Paths The goal is to extend the stereo method of Savinov et al [1] to optical flow. The main challenge is with handling of large memory requirements by passing only restricted subset of most probable labels during the back-propagation phase of label likelihoods. The method could be implemented in any deep learning framework. [1] Savinov et al., “Matching Neural Paths: Transfer from Recognition to Correspondence Search”, NIPS 2017 Required: C++, CUDA, any Deep learning framework Lubor Ladicky, [email protected]

Transfer from Recognition to Optical Flow by Matching Neural … · 2018. 2. 23. · 3D Vision, Spring Semester 2016 Goa l: Requirements / Tools: Supervisor: Descriptio n: Goal: Description:

  • Upload
    others

  • View
    0

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Transfer from Recognition to Optical Flow by Matching Neural … · 2018. 2. 23. · 3D Vision, Spring Semester 2016 Goa l: Requirements / Tools: Supervisor: Descriptio n: Goal: Description:

3D Vision, Spring Semester 2016

Goal:

Requirements / Tools:

Supervisor:

Description:

Goal:

Description:

2018

Transfer from Recognition to Optical Flow by Matching Neural Paths

Implementation of Optical Flow Method by Matching Neural Paths

The goal is to extend the stereo method of Savinov et al [1] to optical flow. The main challenge is with handling of large memory requirements by passing only restricted subset of most probable labels during the back-propagation phase of label likelihoods. The method could be implemented in any deep learning framework.

[1] Savinov et al., “Matching Neural Paths: Transfer from Recognition to Correspondence Search”, NIPS 2017

Required: C++, CUDA, any Deep learning frameworkLubor Ladicky, [email protected]