This project solves the problems of recognizing switches on the railway, taking into account the time of day and weather conditions.
✅On support.
To install the project, just clone the project and install the dependencies from the requirement.txt file
pip install -r requirements.txtTo run, configure the paths for the code/model/inference.ipynb file and run the code for training or forward pass
To view the metric values that the model received during training, view the results folder
During development, the version of CUDA 12.1 for pytorch was used, so the same version is indicated in the requirements file
torch==2.2.1+cu121
torcaudio==2.2.1+cu121
torchvision==0.17.1+cu121
Check if these cuda and cuDNN drivers match
On this drive you can find all the data that was used or generated in this project Google Drive
fe7208e4-5709-4f9f-bd35-6c3e91ac48eb.mp4
TODO:
- Enrich training examples to improve model accuracy
- Play with threshold values
IoUandConf