Master thesis project at Cranfield University in 2016. The goal is to localise and recognise food items on a picture. Combination of image processing and machine learning.
To get more information on previous work, the methodology or the results, see my thesis report in the sub-directory thesis.
Install:
-
Python 3 (version: 3.5.2)
-
Caffe and PyCaffe (version : rc3)
Make sure that caffe can be imported in your project with
import caffe
Then, install using the Model Zoo (and especially the installation using the script included in caffe) these two pre-trained models:
-
OpenCv 3 and its Opencv python (version: 3.1) with opencv_contrib (version 3.1)
Again, make sure that opencv can be imported in your project with
import cv2
-
Optional: create a virtual environment for the project
-
Clone the repository
git clone https://github.com/bnogaret/food_log.git
-
Install python dependencies:
pip install -r requirements.txt
-
Download and unzip in /data/ directory UEC FOOD256 and/or UEC FOOD100. See section 5 of my thesis report for the reasons behind this dataset choice.
โ ๏ธ Be careful, I have modified added / modified some files from these archives (they are included in the repo).
To run the (too few) tests, execute from the root directory:
cd src/tests/
python3 -m unittest discover
To generate the documentation, sphinx must be installed.
pip install sphinx
To compile the documentation, the below command must be executed in the root directory:
sphinx-build -b html docs/ build/