ML Application · Jupyter Notebook
Multi-Task-Rail
A Multi Task Network with 2 heads: one for rail signal regression (ResNet50) and a second for rail signal semantic segmentation (FPN decoder) connected to the GERAL (German Rail Light Signals) and RailSem19 (Semantic Rail Scene Understanding)
What it is
This repository provides a Jupyter notebook implementing a multi-task model for railway signal detection (using GERALD dataset) and scene segmentation (using RailSem19 dataset) with a shared ResNet50-FPN backbone and task-specific heads. The code includes data loaders, model definitions, a training loop, and segmentation inference, but lacks detection inference and has not been executed.
Where it counts
- Computer Vision90%
- Machine Learning80%
- Data Science30%
How much of the project is about each field, as scoring counts it. It ranks in the field it is most about.
Around it on the Computer Vision board
Whole board- 05FootballNetML Application · Jupyter Notebook48.2ImplementationSuceveanu Mircea Dan SanduComputer Vision
- 06background-subtractionEducational Implementation · Jupyter Notebook42.1ImplementationPop AlexandruComputer Vision
- 07Multi-Task-RailML Application · Jupyter Notebook40.1ImplementationSuceveanu Mircea Dan SanduComputer Vision
- 08the-land-of-wolves-redm-visualsEducational Implementation · Lua37.6ExplorationMaster MindComputer Vision
- 09enhanced-negative-promptsDataset / Data Pipeline36.5ExplorationMaster MindComputer Vision
