An end to end under water animal identification machine learning model using Quantum kernels along with Classical pretrained models. This project comprises a brief research on three popular classical models ResNet18 , ResNet50 ,InceptionV3 and its quantum counter parts Hybrid Quantum classical ResNet18 | ResNet50 | InceptionV3.
- To provide a web app backboned by Hybrid quantum classical CNN models
- Hybrid Quantum ML model will use PQC as Quantum kernels along with pretrained classical models
- Whole model will be trained on UAD-2023 dataset containing >13k images for 23 classes
- Entire model is made learnable , quantum kernels will use classical gradient descent for optimization
- Distributing all kernels across multiple cloud QPU's for parallelism
- Using reversible nature of quantum circuits to compute gradients , thus bringing time complexity from exponential to constant
If you want to run inference on real IBM Quantum device you need to add IBM Quantum token to config.toml file
API_KEY
Clone this repo
git clone https://github.com/nagarajRPoojari/Quantum-Threads.gitinstall requirements
pip install -r requirements.txtstart streamlit server
streamlit run main.py
- Tensorflow
- Pennylane
- Qiskit
- Amazon braket
- streamlit
- AWS