- Image streaming between Raspberry Pi and Python server.
- Fire detection & alarm
- Mobile application for uses
- Firebase Real-time database
- Firebase Cloud Messaging(FCM)
-
Maintain & Upgrade this project.
-
Make performance research graphs.
-
Add Smoke detection.
-
Fire & smoke data collecting.
-
Apply latest deep learning model, for example, YOLO v3.
-
Data augmentation.
-
Gather the information
-
Test Demo model on Raspberry Pi 3 B+
-
Make train dataset
-
First train custom model
-
Test model
-
If needed, increase a performance of the model
-
Make server application
(Done: receive JSON data from android)
-
Make client application
(Done: Recycler Popup window, splash, Push alarm, HD, Call 119)
-
System Test
-
Communication between Raspberry Pi and Python server
-
Make a final report and demonstration video
-
Build train enviornment
-
Make up datasets for testing model's accuracy
-
Check clear commumication among Raspberry Pi, Python Server and Android Client
-
Design a user-friendly UI/UX on android client app
-
Make a database server
-
Additional functionality.
(Done: Server recording, and removing oldest file when it is expiring, Getting detection result from server using log)
-
Build on AWS server for demonstration.
-
License validation
-
Function Test (10/31)
-
[DEMO]
Make demo server and client(success connecting python server and android client using TCP socket.)
- The TOD(TensorFlow Object Detection) on the Raspberry Pi run environments are Tensorflow 1.9, cudNN 7.2.1 and cuda 9.0(Those are the best setting without error)
- Firebase library dosen't work in Python 3.7
-
The Tensorflow official repository
-
The method to transplant deep learning model on Raspberry pi 3
-
The method to train deep learning model using tensorflow object detecion API.
-
PyFCM
-
Python-Firebase