Boost Medical Image Classification Accuracy
Budget: $10 – $120 USD
Project Description:
We are hiring a deep learning engineer for a medical image classification project with a strict focus on achieving a significant performance improvement over existing solutions.
This is not a simple replication task. You are expected to:
- Search for and select two top-tier public datasets suitable for classification
- Find the most advanced and highest-performing public codebases or research papers in this field
- Benchmark those solutions and propose an enhancement strategy to exceed their accuracy
- Implement a customized, optimized model that delivers measurable improvement over current state-of-the-art
Your Responsibilities:
- Identify and justify two public datasets relevant to the classification goal
- Analyze the best available code and academic research on similar tasks
- Design and implement enhancements that push the accuracy beyond what those solutions report
Clearly document:
- The sources you reference (papers, datasets, models)
- Your full approach and rationale
- Evaluation metrics and performance comparisons
- Submit clean, modular, and well-documented code + a summary report
Performance Expectations:
- This is a performance-critical task
- You must achieve a significant accuracy improvement over the current best-reported numbers in the field
For example, if the best reported result is 95%, your model should target 96–97% or more
- Replicating existing work is not enough — original enhancement is required
- The project will not be considered complete unless the accuracy gain is achieved and fully documented
Other Requirements:
- All source material must be clearly cited
- All improvements must be original and measurable
- Code should be clean, organized, and reproducible
- A summary of your methodology and results must be included
Strict Confidentiality Policy:
- You may not share or publish this work under any circumstances
- It is not permitted in your CV, GitHub, portfolio, website, or elsewhere
- All rights to the code and documentation will transfer to the client upon delivery
- Any breach of confidentiality will void the project agreement
We are hiring a deep learning engineer for a medical image classification project with a strict focus on achieving a significant performance improvement over existing solutions.
This is not a simple replication task. You are expected to:
- Search for and select two top-tier public datasets suitable for classification
- Find the most advanced and highest-performing public codebases or research papers in this field
- Benchmark those solutions and propose an enhancement strategy to exceed their accuracy
- Implement a customized, optimized model that delivers measurable improvement over current state-of-the-art
Your Responsibilities:
- Identify and justify two public datasets relevant to the classification goal
- Analyze the best available code and academic research on similar tasks
- Design and implement enhancements that push the accuracy beyond what those solutions report
Clearly document:
- The sources you reference (papers, datasets, models)
- Your full approach and rationale
- Evaluation metrics and performance comparisons
- Submit clean, modular, and well-documented code + a summary report
Performance Expectations:
- This is a performance-critical task
- You must achieve a significant accuracy improvement over the current best-reported numbers in the field
For example, if the best reported result is 95%, your model should target 96–97% or more
- Replicating existing work is not enough — original enhancement is required
- The project will not be considered complete unless the accuracy gain is achieved and fully documented
Other Requirements:
- All source material must be clearly cited
- All improvements must be original and measurable
- Code should be clean, organized, and reproducible
- A summary of your methodology and results must be included
Strict Confidentiality Policy:
- You may not share or publish this work under any circumstances
- It is not permitted in your CV, GitHub, portfolio, website, or elsewhere
- All rights to the code and documentation will transfer to the client upon delivery
- Any breach of confidentiality will void the project agreement