Team members:
Have you ever seen a beautiful flower and wondered what kind of flower it is? Have you ever stepped on a pretty leaf and questioned which tree did it possibly fall of? Well, you're not the first In this website, you can get the answer to all your questions by simply uploading a picture!
To build the website, we used Flask, along with Python. Users can upload a picture of any format; an alert will pop-up if it is the wrong format, say a pdf. To train the model, we used to dataset by Maria-Elena Nilsback and Andrew Zisserman and trained the model using different methods in order to get the most accurate results.
We did not get good accuracies at first, therefore we had to train the model using more precise data. Linking the website's back-end to the training model code was time-consuming too, as we were not very familiar in that field.
We finished the project 10min before the deadline, everything working, the website looks nice at the end, and we solved each issue as a team; teamwork energy to the max :) We successfully got an accuracy of more than 67%
We learnt Flask on spot and created a website from scratch. We were able to communicate efficiently even if it was purely virtual.
We are trying to do a basemap in the future too, where when the user uploads the picture, it will be shown on a map where the flower or the leaf was discovered, so that the user will have an idea of the different species present in the world. Would be nice if one day, I travel abroad, see a pretty flower, upload to the website, and have my tracker keeping the info for me.
- Clone the repository
git clone https://github.com/42694647426/MAISHACK2021.git
- Install dependencies
pip install -r requirements.txt
- Run flask
set FLASK_APP='official_website/websiteTrial.py'
flask run
