Skip to content

README.md

Latest commit

 

History

History
executable file
·
70 lines (53 loc) · 3.65 KB

README.md

File metadata and controls

executable file
·
70 lines (53 loc) · 3.65 KB

Deep Learning

https://github.com/ezerilli/Deep_Learning

SETTING UP THE ENVIRONMENT 👨🏻‍💻👨🏻‍💻👨🏻‍💻

The following steps lead to setup the working environment for Deep Learning projects. 👨🏻‍💻‍📚‍‍‍‍

Installing the conda environment is a ready-to-use solution to be able to run python scripts without having to worry about the packages and versions used. Alternatively, you can install each of the packages in requirements.yml on your own independently with pip or conda.

  1. Start by installing Conda for your operating system following the instructions here.

  2. Now install the environment described in requirements.yaml:

conda env create -f requirements.yml
  1. To activate the environment run:
conda activate DL
  1. Once inside the environment, if you want to run a python file, run:
python my_file.py
  1. To deactivate the environment run:
conda deactivate
  1. During the semester I may need to add some new packages to the environment. So, to update it run:
conda env update -f requirements.yml

Fashion-MNIST ‍🔥🔥🔥

This assignment aims to explore the Fashion-MNIST dataset and its applications to clothing detection and recognition in real world images and videos.

The project consists of two parts:

  • experiment 1, training and testing 3 different models (MiniVGG4, MiniVGG6 and MiniVGG6 with Data Augmentation ) and producing corresponding accuracy/loss curves + generate subplot of test images with predictions.

  • experiment 2, clothing detection and recognition in real world images and videos (in progress...).

In order to run the experiments, run:

cd Fashion_MNIST
python run_experiments.py

Figures will show up progressively. It takes a while to perform all the experiments. By default, models training is commented out and the script loads pre-trained weights and then assesses performances on the test set. However, training plots have already been saved into the images directory. Theory, results and experiments are discussed in the report.

REFERENCES

  • [1] Yann LeCun, Léon Bottou, Yoshua Bengio, and Patrick Haffner. Gradient based learning applied to document recognition. IEEE, 1998.
  • [2] Olga Russakovsky, Jia Deng, Hao Su, Jonathan Krause, Sanjeev Satheesh, Sean Ma, Zhiheng Huang, Andrej Karpathy, Aditya Khosla, Michael Bernstein, Alexander C. Berg, and Li Fei-Fei. ImageNet Large Scale Visual Recognition Challenge. International Journal of Computer Vision (IJCV), 2015.
  • [3] Alex Krizhevsky, Ilya Sutskever, and Geoffrey E. Hinton. Imagenet Classification with Deep Convolutional Neural Networks. NIPS, 2012.
  • [4] Karen Simonyan and Andrew Zisserman. Very Deep Convolutional Networks for Large-Scale Image Recognition. arXiv preprint arXiv:1312.6082v4, 2014.
  • [5] Christian Szegedy, Wei Liu, Yangqing Jia, Pierre Sermanet, Scott Reed, Dragomir Anguelov, Dumitru Erhan, Vincent Vanhoucke, and Andrew Rabinovich. Going Deeper with Convolutions. In Computer Vision and Pattern Recognition (CVPR), 2015.
  • [6] Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep Residual Learning for image recognition. arXiv preprint arXiv:1512.03385, 2015.
  • [9] Han Xiao, Kashif Rasul, and Roland Vollgraf. Fashion-MNIST: a novel image dataset for benchmarking machine learning algorithms, 2017.
  • [10] Keras: The Python Deep Learning library. https://keras.io. Last accessed: 2019-10-24.
  • [11] Google Cloud Platform for fast Deep Learning development. https://cloud.google.com/deep-learning-vm/. Last accessed: 2019-10-24.