ATLAS_Coding_Challenge

Changelog - Laryn Qi

Main Task: AtlasAnnotationTool.__init__, AtlasAnnotationTool.topCanvasClicked, and AtlasAnnotationTool.clearSelected in main.py
- Added two instance attributes, selected and sizes, in the constructor to keep track of selected points and their original colors as well as point sizes.
- In AtlasAnnotationTool.topCanvasClicked, we store the a copy of the selected point's original color. Then, its color is changed to be red so that it stands out. Selected points also have their size magnified by 3x.
- Now, whenever the user clears their selected floodfill points (either by clicking "Done" or "Cancel"), the selected points revert back to their original colors and size.
- AtlasAnnotationTool.topCanvasClicked now ignores repeated points.
- Fixed a bug in distance_traveled to handle null inputs
Stretch Goal 1 (Coordinate Crop): main.ui, AtlasAnnotationTool.wireWidgets, AtlasAnnotationTool.setOnClickListener, AtlasAnnotationTool.btn_crop_clicked, and crop_remove (in custom_util.py)
- Using QT Designer, updated the "Your Function" tab in main.ui to contain 3 input fields for the X, Y, and Z coordinates and a "Crop" button. Renamed this tab to "Coordinate Crop"
- Properly linked the new UI elements to the backend in AtlasAnnotationTool.wireWidgets and AtlasAnnotationTool.setOnClickListener.
- AtlasAnnotationTool.btn_crop_clicked checks for malformed inputs and then runs crop_remove and renders the result.
- Modified the crop_remove algorithm to cut out the desired region from the pointcloud.
- To see a demo of this crop functionality, load scene.ply, enter in X = 3877, Y = 105, Z = 5241, and click "Crop"
  - You should see the brown wall being cropped out in the lower canvas.
Stretch Goal 2 (Deletion): prompt_deleting (in custom_util.py) and AtlasAnnotationTool.btn_delete_clicked
- Added new prompt that receives the name of the save/segment the user wishes to delete.
  - Properly handles invalid segment names.
- AtlasAnnotationTool.btn_delete_clicked deletes the save in the frontend and backend.
  - Properly handles reordering the remaining saves.
  - Deletes the .json if no saves remain.

References - Laryn Qi

Open3D Point Cloud for background information
NumPy for edits to crop_remove and array copying for coloring
Qt for Python for making UI changes
VisPy for modifying and displaying the points

I. Get Started

A. Basics

We do expect you to be familiar with Git. Please create your own repository for this coding challenge. You may use git clone https://github.com/wuxiaohua1011/ATLAS_Coding_Challenge.git to get a copy of the skeleton code

B. Set up environment

We use Conda as our environment management tool

conda create -n env_name python=3.7
conda activate env_name
pip install -r requirements.txt
To see if everything is working properly, run python main.py and the following window should pop up

C. Sample Usage

Run the program with python main.py.
click on the Floodfill tab
click on the Load button
Select data/scene.ply
You should see this screen
click on any three points on a wall, and then click Done
you should see something like this

Your Task

Introduction

The goal of this project is to create a software that can help segment and label point clouds for the entire UC Berkeley. The design of this software will be quite complex due to the nature of how big the dataset is and the degree of flexibility that we need to incorporate into the design of this program.

This coding challenge's goal is to

assess your capability to work with ambiguity
assess your capability to work with open source libraries and codes with few documentation
assess your capability to understand existing codebase

Please note that this coding challenge is representative of the work that you will be actually doing in the project.

Specifically, below are the functionalities that we want you to implement:

Any interactions(buttons, textfields, etc) you decide to implement should be in the Your Function tab
Display the original pointcloud in the upper display box(the upper display box with the room in the Sample Usage section)
Display an arbitrary cropped out pointcloud in the lower display box(the lower display box with the wall in the Sample Usage section)
Currently, when you click there are no points showing, but in the backend, it register a point click. We noted where you can display points in main.py, line 239. Implement when canvas is clicked, a point will be displayed functionality

Stretch Goal:

In the Your Function tab, implement input fields that user can provide arbitrary X, Y, Z coordinates, and using those X,Y,Z coordinates, crop out the respective portion of the pointcloud. If any error occurs, write it in the message_center in the lower left corner
Implement function for saving and deleting. (Hint: saving is done for you, please see how we are doing it and make sure that you comply with the same schema)

Deliverable:

Please send us an email at wuxiaohua1011@berkeley.edu with the following content:
1. The Github Repository URL that we can review your coding challenge
2. Your Resume
Please make sure that if you installed any external packages, indicate in the Get Started Section or revise the requirements.txt. You are responsible for making sure that the program runs on any computers, and that the we have no problem running and reviewing your code.
Please make extensive comments in the files on where you revised/added code.

Resources

Here are some of the major libraries that this project depends on. Please look into main.py to understand how these libraries are tied together.

Open3D - http://www.open3d.org/
VisPy - http://vispy.org/
PyQt5 - https://www.riverbankcomputing.com/static/Docs/PyQt5/introduction.html

There are many tutorials/documentation online, feel free to use Google for more documentations.

Miscellaneous

If at any point, you have any questions or concerns, please email us at wuxiaohua1011@berkeley.edu, we'll get back to you ASAP.

LarynQi/ATLAS_Coding_Challenge