Open Circles

Open Circles : Bridging Minds Beyond Borders is an international university collaboration connecting students globally through IT and computer science. This immersive experience features workshops, talks, and activities that inspire, educate, and foster cross-cultural connections, creating a global network of future tech leaders.

OC Archives

Add all past OC projects here, swap the following h2 tag with latest OC documentation

• OC 1.0

• OC 2.0 | UPCOMING!

OC 1.0 - Computer Vision from Convolutional to Transformers

Any code generated with GenAI is marked with explicit docstring comments

Datasets

1. Feral-Cat-Segmentation Dataset

   Download the dataset from Kaggle with your preferred format. We recommend pkl.

   • pickle (0.706020459998399 seconds)

   • parquet (1.2935579999990297 seconds)

   • csv (ran out of memory 😰)

   • jpg (not implemented)

   • jpg + json (not implemented)

   Test the dataset using dataset script

   

2. Dogs vs. Cats

   There is another dataset used to train the transfer learning model. Dataset from Kaggle. Download the train.zip & test1.zip and unzip the files.

   Test the dataset using dataset script

Setup & Run

1. Clone this repo

   $ git clone --depth 1 https://github.com/luhouyang/open_circles.git

2. Create a Python environment (replace with your own directory)

   $ python3.12 -m venv D:\User\test

3. Change your Python interpreter (steps if using VS Code). At the search bar type >Python: Select Interpreter -> select Enter interpreter path -> paste the path used to create env i.e. D:\User\test

4. Restart your IDE

5. Activate your environment at the terminal (replace with your own directory)

   $ D:/User/test/Scripts/activate.bat

   You should see (test) C:\Users\...\open_circles> in the terminal

6. Install packages (for pytorch please change according to your device capabilities)

   IMPORTANT! BEFORE INSTALLING CHECK THAT YOUR ENVIRNOMENT IS ACTIVE i.e. (your_env_name) C:\...

   cd OC01
   pip install polars pillow pycocotools tqdm ipykernel torchinfo einops

   install with caution, as the pytorch version in requirements.txt might not match

   pip install -r requirements.txt
   

   Check the PyTorch Website

   pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu124

Reading Materials

• Hisory of CNN

• LeNet-1 | Original CNN Paper

• LeNet-5 | paper

• AlexNet | paper | PyTorch

• GoogLeNet | paper | PyTorch

• VGGNet | paper | PyTorch

• U-Net | paper | PyTorch

• ResNet | paper | PyTorch

• DenseNet | paper | PyTorch

• MobileNetV1 | paper | MobileNetV2 | paper | PyTorch | MobileNetV3 | paper | PyTorch | MobileNetV4 | paper | Hugging Face

• ConvNeXt | paper | PyTorch

• SETR | paper

Data Preprocessing & Visualization

preprocessing

Feral-Cat-Segmentation

Future Additions TODO

• Move setup documentation to respective folders