Update app.py
Browse files
app.py
CHANGED
|
@@ -5,15 +5,12 @@ from PIL import Image
|
|
| 5 |
import os
|
| 6 |
from typing import Tuple, List, Dict
|
| 7 |
|
| 8 |
-
# Единоразовое определение устройства
|
| 9 |
-
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
| 10 |
-
|
| 11 |
def load_model() -> Tuple[torch.nn.Module, List[str]]:
|
| 12 |
"""
|
| 13 |
-
|
| 14 |
Returns:
|
| 15 |
-
model:
|
| 16 |
-
labels:
|
| 17 |
"""
|
| 18 |
model_path = "skinconvnext_scripted.pt"
|
| 19 |
labels_path = "labels.txt"
|
|
@@ -23,18 +20,18 @@ def load_model() -> Tuple[torch.nn.Module, List[str]]:
|
|
| 23 |
if not os.path.exists(labels_path):
|
| 24 |
raise FileNotFoundError("File labels.txt not found.")
|
| 25 |
|
|
|
|
| 26 |
model = torch.jit.load(model_path, map_location=device)
|
| 27 |
model.eval()
|
| 28 |
-
model.to(device) # Перемещаем модель на устройство сразу
|
| 29 |
|
| 30 |
with open(labels_path, "r") as f:
|
| 31 |
-
labels =
|
| 32 |
|
| 33 |
return model, labels
|
| 34 |
|
| 35 |
model, labels = load_model()
|
| 36 |
|
| 37 |
-
#
|
| 38 |
preprocess = transforms.Compose([
|
| 39 |
transforms.Resize((224, 224)),
|
| 40 |
transforms.ToTensor(),
|
|
@@ -44,48 +41,48 @@ preprocess = transforms.Compose([
|
|
| 44 |
|
| 45 |
def predict(image: Image.Image) -> Dict[str, float]:
|
| 46 |
"""
|
| 47 |
-
|
| 48 |
Args:
|
| 49 |
-
image (PIL.Image):
|
| 50 |
|
| 51 |
Returns:
|
| 52 |
-
Dict[str, float]:
|
| 53 |
"""
|
| 54 |
try:
|
| 55 |
-
# Приводим изображение к RGB и преобразуем
|
| 56 |
image = image.convert("RGB")
|
| 57 |
-
image_tensor = preprocess(image).unsqueeze(0)
|
|
|
|
|
|
|
|
|
|
| 58 |
|
| 59 |
with torch.no_grad():
|
| 60 |
output = model(image_tensor)
|
| 61 |
-
# Предполагаем, что output имеет размерность [1, N]
|
| 62 |
scores = torch.nn.functional.softmax(output[0], dim=0)
|
| 63 |
|
| 64 |
-
# Формируем словарь с предсказаниями
|
| 65 |
predictions = {label: float(score) for label, score in zip(labels, scores)}
|
| 66 |
sorted_predictions = dict(sorted(predictions.items(), key=lambda item: item[1], reverse=True))
|
|
|
|
| 67 |
return sorted_predictions
|
| 68 |
except Exception as e:
|
| 69 |
return {"error": str(e)}
|
| 70 |
|
| 71 |
-
# Описание интерфейса Gradio
|
| 72 |
title = "🔥 Skin-AI"
|
| 73 |
description = (
|
| 74 |
"🔬 **Skin-AI — AI-Powered Skin Disease Classification**\n\n"
|
| 75 |
-
"
|
| 76 |
-
"### 🚀
|
| 77 |
-
"1️⃣
|
| 78 |
-
"2️⃣
|
| 79 |
-
"3️⃣
|
| 80 |
-
"⚠️ **
|
| 81 |
-
"### 🛠
|
| 82 |
-
"- PyTorch\n"
|
| 83 |
"- Gradio\n"
|
| 84 |
"- Hugging Face Spaces\n\n"
|
| 85 |
-
"🔗
|
| 86 |
)
|
| 87 |
|
| 88 |
-
#
|
| 89 |
examples = [
|
| 90 |
["example1.jpg"],
|
| 91 |
["example2.jpg"]
|
|
|
|
| 5 |
import os
|
| 6 |
from typing import Tuple, List, Dict
|
| 7 |
|
|
|
|
|
|
|
|
|
|
| 8 |
def load_model() -> Tuple[torch.nn.Module, List[str]]:
|
| 9 |
"""
|
| 10 |
+
Loads the model and class labels.
|
| 11 |
Returns:
|
| 12 |
+
model: The loaded PyTorch model.
|
| 13 |
+
labels: List of class labels.
|
| 14 |
"""
|
| 15 |
model_path = "skinconvnext_scripted.pt"
|
| 16 |
labels_path = "labels.txt"
|
|
|
|
| 20 |
if not os.path.exists(labels_path):
|
| 21 |
raise FileNotFoundError("File labels.txt not found.")
|
| 22 |
|
| 23 |
+
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
| 24 |
model = torch.jit.load(model_path, map_location=device)
|
| 25 |
model.eval()
|
|
|
|
| 26 |
|
| 27 |
with open(labels_path, "r") as f:
|
| 28 |
+
labels = [line.strip() for line in f.readlines()]
|
| 29 |
|
| 30 |
return model, labels
|
| 31 |
|
| 32 |
model, labels = load_model()
|
| 33 |
|
| 34 |
+
# Define image preprocessing steps
|
| 35 |
preprocess = transforms.Compose([
|
| 36 |
transforms.Resize((224, 224)),
|
| 37 |
transforms.ToTensor(),
|
|
|
|
| 41 |
|
| 42 |
def predict(image: Image.Image) -> Dict[str, float]:
|
| 43 |
"""
|
| 44 |
+
Makes a prediction for the given image.
|
| 45 |
Args:
|
| 46 |
+
image (PIL.Image): The input image.
|
| 47 |
|
| 48 |
Returns:
|
| 49 |
+
Dict[str, float]: A dictionary where keys are class names, and values are probabilities.
|
| 50 |
"""
|
| 51 |
try:
|
|
|
|
| 52 |
image = image.convert("RGB")
|
| 53 |
+
image_tensor = preprocess(image).unsqueeze(0)
|
| 54 |
+
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
| 55 |
+
image_tensor = image_tensor.to(device)
|
| 56 |
+
model.to(device)
|
| 57 |
|
| 58 |
with torch.no_grad():
|
| 59 |
output = model(image_tensor)
|
|
|
|
| 60 |
scores = torch.nn.functional.softmax(output[0], dim=0)
|
| 61 |
|
|
|
|
| 62 |
predictions = {label: float(score) for label, score in zip(labels, scores)}
|
| 63 |
sorted_predictions = dict(sorted(predictions.items(), key=lambda item: item[1], reverse=True))
|
| 64 |
+
|
| 65 |
return sorted_predictions
|
| 66 |
except Exception as e:
|
| 67 |
return {"error": str(e)}
|
| 68 |
|
|
|
|
| 69 |
title = "🔥 Skin-AI"
|
| 70 |
description = (
|
| 71 |
"🔬 **Skin-AI — AI-Powered Skin Disease Classification**\n\n"
|
| 72 |
+
"This project utilizes a deep learning model to classify skin diseases based on an uploaded image.\n\n"
|
| 73 |
+
"### 🚀 How to Use:\n\n"
|
| 74 |
+
"1️⃣ Upload or take a photo of the affected skin area.\n\n"
|
| 75 |
+
"2️⃣ Click the 'Submit' button.\n\n"
|
| 76 |
+
"3️⃣ The app will return the probabilities for possible skin conditions.\n\n"
|
| 77 |
+
"⚠️ **Important!** The results are for informational purposes only and do not constitute a medical diagnosis.\n\n"
|
| 78 |
+
"### 🛠 Technologies Used:\n"
|
| 79 |
+
"- PyTorch (Lightning)\n"
|
| 80 |
"- Gradio\n"
|
| 81 |
"- Hugging Face Spaces\n\n"
|
| 82 |
+
"🔗 Source Code: [Hugging Face](https://huggingface.co/Eraly-ml/Skin-AI)"
|
| 83 |
)
|
| 84 |
|
| 85 |
+
# Adding example images
|
| 86 |
examples = [
|
| 87 |
["example1.jpg"],
|
| 88 |
["example2.jpg"]
|