Upload 4 files
Browse files- README.md +41 -20
- best_model.h5 +3 -0
- local_finetuned_model.h5 +3 -0
- malaria_model.tflite +3 -0
README.md
CHANGED
|
@@ -1,20 +1,41 @@
|
|
| 1 |
-
-
|
| 2 |
-
|
| 3 |
-
|
| 4 |
-
-
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
-
|
| 8 |
-
-
|
| 9 |
-
-
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
|
| 19 |
-
|
| 20 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# AI Malaria Detection System - Ghana Model
|
| 2 |
+
|
| 3 |
+
## Overview
|
| 4 |
+
This system uses deep learning (MobileNetV2) to detect malaria parasites in blood smear images. The model has been fine-tuned specifically on Ghanaian blood smear samples to ensure accuracy with local staining and imaging conditions.
|
| 5 |
+
|
| 6 |
+
## Performance Metrics
|
| 7 |
+
- **Overall Accuracy**: 97.2%
|
| 8 |
+
- **Parasitized Detection**: 97.8%
|
| 9 |
+
- **Uninfected Rejection**: 92.3%
|
| 10 |
+
- **Model Size**: 2.54 MB (mobile-optimized)
|
| 11 |
+
|
| 12 |
+
## Folder Structure
|
| 13 |
+
```
|
| 14 |
+
/malaria_detector/
|
| 15 |
+
βββ models/
|
| 16 |
+
β βββ local_finetuned_model.h5
|
| 17 |
+
β βββ malaria_model.tflite
|
| 18 |
+
βββ scripts/
|
| 19 |
+
β βββ predict_patches.py
|
| 20 |
+
β βββ requirements.txt
|
| 21 |
+
βββ evaluation/
|
| 22 |
+
β βββ final_report.pdf
|
| 23 |
+
β βββ confusion_matrix.png
|
| 24 |
+
βββ README.md
|
| 25 |
+
```
|
| 26 |
+
|
| 27 |
+
## Deployment & Usage
|
| 28 |
+
1. **Clone the Repository**: Download the entire `malaria_detector` folder.
|
| 29 |
+
2. **Install Dependencies**: Navigate to the `scripts/` folder and install required libraries:
|
| 30 |
+
```bash
|
| 31 |
+
pip install -r requirements.txt
|
| 32 |
+
```
|
| 33 |
+
3. **Model Location**: Ensure `local_finetuned_model.h5` and `malaria_model.tflite` are in the `models/` folder.
|
| 34 |
+
4. **Run Prediction**: Use the `predict_patches.py` script to analyze whole slide images:
|
| 35 |
+
```bash
|
| 36 |
+
python scripts/predict_patches.py /path/to/your/image.jpg [confidence_threshold]
|
| 37 |
+
```
|
| 38 |
+
* `[confidence_threshold]` is optional; defaults to `0.5` (50%). Recommended: `0.5`.
|
| 39 |
+
|
| 40 |
+
## Contact
|
| 41 |
+
For questions or further assistance, please contact the development team.
|
best_model.h5
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:78a322f1814d483654abca7fe053aef096e8d41a0c614fbc266fb9532c7cb9f2
|
| 3 |
+
size 11540648
|
local_finetuned_model.h5
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:177f02f505c4384b1dd11a44ac66c1682357e4aa3e448e0555ba6afcc6dd974d
|
| 3 |
+
size 17305376
|
malaria_model.tflite
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fe24b179408842b3647ad02dfecdc491db070e1d2fc5cb77ddad5342b3b46c1e
|
| 3 |
+
size 2667336
|