Chunlan Wang,M.Arch

Master’s Student
School of Architecture
Georgia Institute of Technology

Email:wchunlan07@gmail.com
Linkedin

Academic Works
Computational Design
HCI/DM
Architecture

Professional Projects
Architecture

Publications

About

CV

Portfolio


Fragment Collection Game Thu Jul 19 2022
Team:Chunlan Wang, Miao Zhang,
Hanwei Zhang,Yiqu Zhang
Instructor:DÉBORAH LÓPEZ, HADIN CHARBEL
Site:Arctic/Hongkong Period:10/2020-01/2021
Software:Yolo system+Blender +Unity/Grasshoper/Photoshop
We use YOLO for real-time object detection, it is fast but the recognition is based on bounding boxes, the grid may contain multiple objects, human participation is needed to get better accuracy.
In our game, player collect the virtual fragments to conjure the landscape. At the same time, object detection and HP monitoring in the sentient system can be further improved with the help of reCAPTCHA and ecological simu- lation in the game.First, the user scans in the city, and required to select all grids that objects are in, then get fragments with ecological information.The results of player's box selection will be used as training set&to make object detection better trained.



We use YOLO for real-time object detection, it is fast but the recognition is based on bounding boxes, the grid may contain multiple objects, human participation is needed to get better accuracy.


So for our object detection footages, we will have a Gradient for the difficulty of recognition, according to this rule, we project these virtual footages in the city.



In our game, player collect the virtual fragments to conjure the landscape. At the same time, object detection and HP monitoring in the sentient system can be further improved with the help of reCAPTCHA and ecological simulation in the game.




 the user scans in the city, and required to select all grids that objects are in, then get fragments with ecological information.



The results of player's box selection will be used as training set，to make object detection better trained.



The results of player's box selection will be used as training set，to make object detection better trained. 







When you gathered the required fragments of the landscape can begin to conjure. You can choose any location you want.



Now the Conjuring is complete.


After conjuring the landscape, you can continue to add fragments for ecological simulation. The game evaluates HP changes based on existing archive data. Each player's modifications  are recorded, in effect helping us to find the optimal HP value. Imagine being in a real Arctic autonomous region, These high HP correspond to areas, for example a certain sea area, which can be used for production and exchange. 
©Chunlan WangPortfolio2019-2024
Return   Next →