mosquito-monitoring

Identify mosquito species and collect data to anticipate epidemics propagation for a CentraleSupelec project

Deployed app (old) : mosquito-monitor.herokuapp.com
Deployed app (our own server with persistent database, temporary url) : mosquito.paulasquin.com

Introduction

Project purpose

This project aims to monitor and later anticipate mosquito related epidemics. Crossing the information about mosquito species, localization, temperature and population density may be really powerful to intuite the evolution or the apparition of an epidemic. Thus, we want to create a plateform where anyone can participate to this big-scale survey.

The user can upload a mosquito picture and will be asked to provide additional information as they localization or the mosquito species if known. Those information will be written in a database, and a Machine Learning algorithm will try to identify where the mosquito is located in the image and what is its species.

Project graph

We initiated a Dockerfile and a docker-compose.yml file in order to wrap all the requirements and dependancies of this project.

You have then 2 choices:

• The 1st one is to install all the requirements (see Installation below)
• The second one is to have Docker (version 1.13.0+ and above) installed and running on your OS

Installation (only if you are not using Docker)

NB: You need to have python 3.6 installed and not 3.7 installed on your machine.

git clone https://github.com/mosquito-boys/mosquito-monitoring.git
cd mosquito-monitoring
pip3 install -r requirements.txt

You may have to install some additional libraries for opencv.

Start server

The server will listen on port 5000. You will need a Google Vision API Key written in the .env file. You will have to create this file at the root of the project, like :

#.env
GOOGLE_APPLICATION_CREDENTIALS=YOUR_KEY

With Docker

Be sure to have docker and docker-compose installed. If you haven't them already, you can follow those tutorials for docker and docker-compose.
Then run

git clone https://github.com/mosquito-boys/mosquito-monitoring.git
cd mosquito-monitoring
# You should copy in this directory the .env file with your API keys, then run
sudo docker-compose up

If you want to rebuild the project you can run

docker-compose build

With Docker on a server

If you want to start the project on your server startup, you can follow this procedure:

cd ~
# Working in personal folder
git clone https://github.com/mosquito-boys/mosquito-monitoring.git
cd mosquito-monitoring

You should copy in this directory the .env file with your API keys.
Also, in order to enable the systemctl service, you will have to edit the docker-mosquito.service. WorkingDirectory=/home/USERNAME/mosquito-monitoring by replacing USERNAME with your server username.
Note : we didn't indicate to use /var/www/ as the .env may be exposed. Then run:

sudo docker-compose -f docker-compose-ssl.yml build
sudo cp docker-mosquito.service /etc/systemd/system
sudo systemctl enable docker-mosquito
sudo systemctl restart docker-mosquito

This last command will run docker-compose -f docker-compose-ssl.yml up

Note: the project support ssl certificates! You can edit docker-compose-ssl.yml to mount your own certificates.
Ours were generated with Let's Encrypt

Without Docker

Debug mode

FLASK_APP=server.py FLASK_DEBUG=1 python3 -m flask run

Normal mode

python3 server.py

Label a mosquito

Import the mosquito classification module

import classification.command_classification as command_classification

Request the labelling

command_classification.label_automatic(path_img)

Return example

[['aedes', '0.8780854'], ['culex', '0.11636846'], ['anopheles', '0.0055461014']]

Additional commands

Remove the database (useful for development purpose)

rm ./sqlite.db

Freeze local packages into requirements.txt with following command

pipreqs ./ --force      

Run the tests

Don't run directly the scripts independently. To test them, run the corresponding test script like below:

python3 -m tests.my_test

Retrain

To run a retrain of Inception

python3 -m tests.test_command_classification --retrain

You shall now wait for 3 to 7 minutes depending on your CPU

Prediction

To run an image labelling test

python3 -m tests.test_command_classification --label [optional path to one or more images]

Ouput :

Testing image labelling
/home/paul/Projects/POOA/mosquito-monitoring/classification/tensorflow
['/home/paul/Projects/POOA/mosquito-monitoring/classification/tensorflow/export_1/graph.db']
['/home/paul/Projects/POOA/mosquito-monitoring/classification/tensorflow/export_1/labels.txt']
['/home/paul/Projects/POOA/mosquito-monitoring/classification/tensorflow/export_1/cmd.txt']
python3 /home/paul/Projects/POOA/mosquito-monitoring/classification/label_image.py --graph=/home/paul/Projects/POOA/mosquito-monitoring/classification/tensorflow/export_1/graph.db --labels /home/paul/Projects/POOA/mosquito-monitoring/classification/tensorflow/export_1/labels.txt --input_layer=Placeholder --output_layer=final_result --image /home/paul/Projects/POOA/mosquito-monitoring/dataset/aedes/pic_001.jpg
2018-10-23 14:28:34.619292: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
[['aedes', '0.8780854'], ['culex', '0.11636846'], ['anopheles', '0.0055461014']]

Project Slides