A modified facade recognization system including AR application to adding advertisement overlay.
- Facade Recognition: Python 3.5, Tensorflow, Opencv, MaskRCNN, matplotlib, yaml, PIL, skimage, datetime.
- Tool: jupyter.
- AR: Unity.
- Clone Repository
git clone https://github.com/raphealshirley/Facade-Proposal.git
- Run server:
Under FacadeProposalfARApplication/server.
Noted that different platforms should be under same loacal area network and do not forget to change IP address in the file.
python3 ./TcpServer.py
- Run ML model under Facade Proposal folder.
python3 ./facade.py
Modified MaskRCNN improved facade recognition recall by 11.8% compared with SegNet using Tensorflow.
Model training and testing on SageMaker; AR application;
Design Web UI; Deployed on AWS.
This system is licensed under MIT license.