Comprehensive HC18 Dataset and Model Development
Budget: ₹12,500 – ₹37,500 INR
1) Data Handling: Organize HC18 dataset (train/test), preprocess DICOM → PNG/JPG, handle
missing masks, integrate more datasets later.
•2) Model Development: Implement and train deep learning pipeline (PyTorch/TensorFlow), store
checkpoints.
•3) Inference Pipeline: Fix errors in inference.py & model.py, enable DICOM input → measurement
output.
•4) Logging & Monitoring: Replace print() with logging, add error handling.
•5) Testing: Create unit tests in tests/ folder, run smoke tests with dummy images.
•6) Deployment: Build REST API (Flask/FastAPI) for hospital uploads, optional frontend
(React/Streamlit), Docker support.
•7) Documentation: Provide README.md, installation guide, requirements.txt, usage examples.
•8) Automation: Include train.py, evaluate.py, inference.py; optional Jupyter notebooks for
experiments.
•9) Milestone Delivery: Ensure progress checkpoints for dataset prep, training, inference,
deployment.
10) Export Results Save outputs as PNG + structured CSV/PDF report
* Single command execution: like python inference.py --input sample_image.dcm --output
result.png
do first only one thing Data Handling: Organize HC18(available online) dataset (train/test), preprocess DICOM → PNG/JPG, handle missing masks,
first do this before I pay advance as demo of your service so that I can believe (trust) on you, how your work suits me don't integrate more datasets
missing masks, integrate more datasets later.
•2) Model Development: Implement and train deep learning pipeline (PyTorch/TensorFlow), store
checkpoints.
•3) Inference Pipeline: Fix errors in inference.py & model.py, enable DICOM input → measurement
output.
•4) Logging & Monitoring: Replace print() with logging, add error handling.
•5) Testing: Create unit tests in tests/ folder, run smoke tests with dummy images.
•6) Deployment: Build REST API (Flask/FastAPI) for hospital uploads, optional frontend
(React/Streamlit), Docker support.
•7) Documentation: Provide README.md, installation guide, requirements.txt, usage examples.
•8) Automation: Include train.py, evaluate.py, inference.py; optional Jupyter notebooks for
experiments.
•9) Milestone Delivery: Ensure progress checkpoints for dataset prep, training, inference,
deployment.
10) Export Results Save outputs as PNG + structured CSV/PDF report
* Single command execution: like python inference.py --input sample_image.dcm --output
result.png
do first only one thing Data Handling: Organize HC18(available online) dataset (train/test), preprocess DICOM → PNG/JPG, handle missing masks,
first do this before I pay advance as demo of your service so that I can believe (trust) on you, how your work suits me don't integrate more datasets