Surgicare

Surgicare (Surgical + Care)

SurgiCare is an AI system designed to support post-surgery patient recovery. In this repository, we focus on a wound classification model trained on an open-source dataset. Our objective is to improve the accuracy of wound detection and guide patients in managing their wound recovery efficiently.

• Online Demo: https://surgicare-demo.streamlit.app/

• Wound dataset: https://www.kaggle.com/datasets/ibrahimfateen/wound-classification

• Model Hub: https://huggingface.co/PogusTheWhisper/SurgiCare

• Pretrained Models:

  • Surgicare-V1-large-turbo: https://huggingface.co/PogusTheWhisper/SurgiCare/resolve/main/SurgiCare-V1-large-turbo.keras
  • Surgicare-V1-large: https://huggingface.co/PogusTheWhisper/SurgiCare/resolve/main/SurgiCare-V1-large.keras
  • Surgicare-V1-medium: https://huggingface.co/PogusTheWhisper/SurgiCare/resolve/main/SurgiCare-V1-medium.keras
  • Surgicare-V1-small: https://huggingface.co/PogusTheWhisper/SurgiCare/resolve/main/SurgiCare-V1-small.keras

---

Result of standard models

EfficientnetV2 B3

• Accuracy: 0.6884

Efficientnet B3

• Accuracy: 0.7436

MobileNetV3Large

• Accuracy: 0.6164

MobileNetV3Small

• Accuracy: 0.6199

---

Result of our models!!

EfficientnetV2 B3

• Accuracy: 0.9127
• Training Details: I used EfficientNet-B3 to train for 50 epochs, monitoring the validation loss.

Efficientnet B3

• Accuracy: 0.9062
• Training Details: I used EfficientNet-B3 to train for 25 epochs, monitoring the validation loss.

MobileNetV3Large

• Accuracy: 0.7969
• Training Details: I used MobileNetV3Large to train for 50 epochs, monitoring the validation loss.

MobileNetV3Small

• Accuracy: 0.7812
• Training Details: I used MobileNetV3Small to train for 50 epochs, monitoring the validation loss.