SpotLab/Chagas_ssd

T. cruzi Detection Model

Overview

This repository contains an object detection model trained using the TensorFlow Object Detection API for detecting Trypanosoma cruzi parasites in microscopic images. The model is based on SSD MobileNet V2 architecture and has been developed by Spotlab.

This model is licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License (CC BY-NC-SA 4.0).

Model ID: 2xmn544x

Model Details

• Architecture: SSD MobileNet V2
• Training data: published on zenodo

  Sample type	Training Image	Training Label	Validation Image	Validation Label
  Human CSF	261	512	68	191
  Human Blood thick	61	55	26	14
  Human Blood thin	156	95	154	64
  Mice Blood thin	570	2648	105	503
  Total	1048	3310	353	772

• Performance:

  Metrics	Human	Mice
  Precision	86	95.8
  Recall	87	85
  F1 score	86.5	90.1

Usage Instructions

Prepare the app

1. Download huggingSpot from google play store
2. Download model using model URL and API KEY

3. 

Image Preparation

1. Increase the zoom until the inner square is clearly visible in the preview
2. Once the inner square appears, the model is ready for detection

Example Images

Here are examples of properly prepared images for detection:

)

Example Predictions

The model can detect T. cruzi parasites with high accuracy: