This project is based on Convolutional and Recurrent neural networks in order to generate particle data, detect particles, and finally forecast their future movements.
- Generate synthetic data.
- Built CNN and CNN-LSTM models to detect particle movements.
- Built LSTM model (2 LSTMs of 128 and 64 units) to forecast the particle position, achieved MSE = 0.24 and 0.18 for X and Y coordinates respectively.
- Python Version: 3.7
- CUDA Toolkit 11.5.119
- Packages and Libraries: pandas, numpy, matplotlib, sklearn, keras, math.
Particle Detection:
CNN Model | CNN-LSTM Model |
---|---|
LSTM model:
Forecasting Results:
X-coordinates | Y-coordinates |
---|---|