message two - grasseau/HAhRD GitHub Wiki

Hi all,

I want to emphasize several points:

  1. Objective of our proposal is to transform matterport implementation (2D image) in a 3D implementation ? Is it clear ?
  2. Based on this objective, build a project with a planning, YOUR planning, to reach this goal
  3. Be sure to reach YOUR plan, because Google takes care that you succeed to be paid.
  4. It is better to propose a schedule, a plan less ambitious (even if it is a small part of the tasks proposed in our wiki) which succeed than proposing all tasks in YOUR schedule.
  5. Some time it is difficult to estimate the time to code and DEBUG part of a program, so plan a long period to be sure and at the end of YOUR proposal, write optional tasks

For this project, you must have good skills or knowledge in :

  • Keras and TensorFlow
  • Debugging with Keras and TF
  • The architecture of Mask RCNN (matterport implementation) or Object Detection, .. should not be a problem ... some papers on the Internet

The Evaluation test is coming soon ... it will be on a module of matterport implementation (re-implement it in 3D ...).

Think your are an engineer which have a professional task.

Thank you for your carefully reading,