import os
import numpy as np
import torch
import torch.nn as nn
import gradio as gr
from torchvision.models import efficientnet_v2_m, EfficientNet_V2_M_Weights
import torch.nn.functional as F
from torchvision import transforms
from PIL import Image
#from data_manager import get_dog_description
dog_breeds = ["Afghan_Hound(阿富汗獵犬)", "African_Hunting_Dog(非洲野犬)", "Airedale(艾爾谷犬)",
"American_Staffordshire_Terrier(美國斯塔福郡梗)", "Appenzeller(亞賓澤爾犬)",
"Australian_Terrier(澳大利亞梗)", "Bedlington_Terrier(貝德靈頓梗)",
"Bernese_Mountain_Dog(伯恩山犬)", "Blenheim_Spaniel(布萊尼姆獵犬)",
"Border_Collie(邊境牧羊犬)", "Border_Terrier(邊境梗)", "Boston_Bull(波士頓梗)",
"Bouvier_Des_Flandres(法蘭德斯牧羊犬)", "Brabancon_Griffon(布魯塞爾格里芬犬)",
"Brittany_Spaniel(布列塔尼獵犬)", "Cardigan(卡迪根威爾士柯基犬)",
"Chesapeake_Bay_Retriever(切薩皮克灣獵犬)", "Chihuahua(吉娃娃)",
"Dandie_Dinmont(丹第丁蒙梗)", "Doberman(杜賓犬)", "English_Foxhound(英國獵狐犬)",
"English_Setter(英國雪達犬)", "English_Springer(英國跳獵犬)",
"EntleBucher(恩特雷布赫山地犬)", "Eskimo_Dog(愛斯基摩犬)", "French_Bulldog(法國鬥牛犬)",
"German_Shepherd(德國牧羊犬)", "German_Short-Haired_Pointer(德國短毛指示犬)",
"Gordon_Setter(戈登雪達犬)", "Great_Dane(大丹犬)", "Great_Pyrenees(大白熊犬)",
"Greater_Swiss_Mountain_Dog(大瑞士山地犬)", "Ibizan_Hound(依比沙獵犬)",
"Irish_Setter(愛爾蘭雪達犬)", "Irish_Terrier(愛爾蘭梗)",
"Irish_Water_Spaniel(愛爾蘭水獵犬)", "Irish_Wolfhound(愛爾蘭獵狼犬)",
"Italian_Greyhound(義大利灰狗)", "Japanese_Spaniel(日本狆)",
"Kerry_Blue_Terrier(凱利藍梗)", "Labrador_Retriever(拉布拉多尋回犬)",
"Lakeland_Terrier(湖畔梗)", "Leonberg(獅毛狗)", "Lhasa(拉薩犬)",
"Maltese_Dog(馬爾濟斯犬)", "Mexican_Hairless(墨西哥無毛犬)", "Newfoundland(紐芬蘭犬)",
"Norfolk_Terrier(諾福克梗)", "Norwegian_Elkhound(挪威獵麋犬)",
"Norwich_Terrier(諾利治梗)", "Old_English_Sheepdog(古代英國牧羊犬)",
"Pekinese(北京犬)", "Pembroke(威爾士柯基犬)", "Pomeranian(博美犬)",
"Rhodesian_Ridgeback(羅得西亞脊背犬)", "Rottweiler(羅威納犬)",
"Saint_Bernard(聖伯納犬)", "Saluki(薩路基獵犬)", "Samoyed(薩摩耶犬)",
"Scotch_Terrier(蘇格蘭梗)", "Scottish_Deerhound(蘇格蘭獵鹿犬)",
"Sealyham_Terrier(錫利哈姆梗)", "Shetland_Sheepdog(設得蘭牧羊犬)",
"Shih-Tzu(西施犬)", "Siberian_Husky(西伯利亞哈士奇)",
"Staffordshire_Bullterrier(斯塔福郡鬥牛梗)", "Sussex_Spaniel(蘇塞克斯獵犬)",
"Tibetan_Mastiff(藏獒)", "Tibetan_Terrier(西藏梗)", "Walker_Hound(沃克獵犬)",
"Weimaraner(威瑪犬)", "Welsh_Springer_Spaniel(威爾士跳獵犬)",
"West_Highland_White_Terrier(西高地白梗)", "Yorkshire_Terrier(約克夏梗)",
"Affenpinscher(猴犬)", "Basenji(巴辛吉犬)", "Basset(巴吉度獵犬)", "Beagle(比格犬)",
"Black-and-Tan_Coonhound(黑褐獵浣熊犬)", "Bloodhound(尋血獵犬)",
"Bluetick(布魯提克獵犬)", "Borzoi(俄羅斯獵狼犬)", "Boxer(拳師犬)", "Briard(布里亞犬)",
"Bull_Mastiff(獒犬)", "Cairn(凱恩梗)", "Chow(鬆獅犬)", "Clumber(克倫伯獵犬)",
"Cocker_Spaniel(可卡獵犬)", "Collie(柯利牧羊犬)", "Curly-Coated_Retriever(捲毛尋回犬)",
"Dhole(豺)", "Dingo(澳洲野犬)", "Flat-Coated_Retriever(平毛尋回犬)",
"Giant_Schnauzer(大型雪納瑞犬)", "Golden_Retriever(黃金獵犬)",
"Groenendael(比利時牧羊犬)", "Keeshond(荷蘭毛獅犬)", "Kelpie(澳洲卡爾比犬)",
"Komondor(匈牙利牧羊犬)", "Kuvasz(庫瓦茲犬)", "Malamute(阿拉斯加雪橇犬)",
"Malinois(比利時瑪利諾犬)", "Miniature_Pinscher(迷你杜賓犬)",
"Miniature_Poodle(迷你貴賓犬)", "Miniature_Schnauzer(迷你雪納瑞犬)",
"Otterhound(水獺獵犬)", "Papillon(蝴蝶犬)", "Pug(巴哥犬)", "Redbone(紅骨獵浣熊犬)",
"Schipperke(舒柏奇犬)", "Silky_Terrier(絲毛梗)",
"Soft-Coated_Wheaten_Terrier(愛爾蘭軟毛梗)", "Standard_Poodle(標準貴賓犬)",
"Standard_Schnauzer(標準雪納瑞犬)", "Toy_Poodle(玩具貴賓犬)", "Toy_Terrier(玩具梗)",
"Vizsla(維茲拉犬)", "Whippet(惠比特犬)", "Wire-Haired_Fox_Terrier(硬毛獵狐梗)"]
class MultiHeadAttention(nn.Module):
def __init__(self, in_dim, num_heads=8):
super().__init__()
self.num_heads = num_heads
self.head_dim = max(1, in_dim // num_heads)
self.scaled_dim = self.head_dim * num_heads
self.fc_in = nn.Linear(in_dim, self.scaled_dim)
self.query = nn.Linear(self.scaled_dim, self.scaled_dim)
self.key = nn.Linear(self.scaled_dim, self.scaled_dim)
self.value = nn.Linear(self.scaled_dim, self.scaled_dim)
self.fc_out = nn.Linear(self.scaled_dim, in_dim)
def forward(self, x):
N = x.shape[0]
x = self.fc_in(x)
q = self.query(x).view(N, self.num_heads, self.head_dim)
k = self.key(x).view(N, self.num_heads, self.head_dim)
v = self.value(x).view(N, self.num_heads, self.head_dim)
energy = torch.einsum("nqd,nkd->nqk", [q, k])
attention = F.softmax(energy / (self.head_dim ** 0.5), dim=2)
out = torch.einsum("nqk,nvd->nqd", [attention, v])
out = out.reshape(N, self.scaled_dim)
out = self.fc_out(out)
return out
class BaseModel(nn.Module):
def __init__(self, num_classes, device='cuda' if torch.cuda.is_available() else 'cpu'):
super().__init__()
self.device = device
self.backbone = efficientnet_v2_m(weights=EfficientNet_V2_M_Weights.IMAGENET1K_V1)
self.feature_dim = self.backbone.classifier[1].in_features
self.backbone.classifier = nn.Identity()
# 動態計算 num_heads
self.num_heads = max(1, min(8, self.feature_dim // 64))
self.attention = MultiHeadAttention(self.feature_dim, num_heads=self.num_heads)
self.classifier = nn.Sequential(
nn.LayerNorm(self.feature_dim),
nn.Dropout(0.3),
nn.Linear(self.feature_dim, num_classes)
)
self.to(device)
def forward(self, x):
x = x.to(self.device)
features = self.backbone(x)
attended_features = self.attention(features)
logits = self.classifier(attended_features)
return logits, attended_features
num_classes = 120
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
model = BaseModel(num_classes=num_classes, device=device)
checkpoint = torch.load('/content/drive/Othercomputers/我的 MacBook Pro/Learning/Cats_Dogs_Detector/best_model/best_model_81_dog.pth', map_location=torch.device('cpu'))
model.load_state_dict(checkpoint['model_state_dict'])
# 將模型設置為評估模式
model.eval()
# Image preprocessing function
def preprocess_image(image):
# 如果圖片是 numpy.ndarray 轉換為 PIL.Image
if isinstance(image, np.ndarray):
image = Image.fromarray(image)
# 使用 torchvision.transforms 進行預處理
transform = transforms.Compose([
transforms.Resize((224, 224)),
transforms.ToTensor(),
transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
])
return transform(image).unsqueeze(0)
def predict(image):
try:
image_tensor = preprocess_image(image)
with torch.no_grad():
logits, _ = model(image_tensor)
_, predicted = torch.max(logits, 1)
breed = dog_breeds[predicted.item()] # Map label to breed name
# Retrieve breed description
description = get_dog_description(breed)
# Formatting the description for better display
if isinstance(description, dict):
description_str = f"**Breed**: {description['Breed']}\n\n"
description_str += f"**Size**: {description['Size']}\n\n"
description_str += f"**Lifespan**: {description['Lifespan']}\n\n"
description_str += f"**Temperament**: {description['Temperament']}\n\n"
description_str += f"**Care Level**: {description['Care Level']}\n\n"
description_str += f"**Good with Children**: {description['Good with Children']}\n\n"
description_str += f"**Exercise Needs**: {description['Exercise Needs']}\n\n"
description_str += f"**Grooming Needs**: {description['Grooming Needs']}\n\n"
description_str += f"**Description**: {description['Description']}\n\n"
else:
description_str = description
return description_str
except Exception as e:
return f"An error occurred: {e}"
iface = gr.Interface(
fn=predict,
inputs=gr.Image(label="Upload an image", type="numpy"), # 支持拖放和圖片編輯
outputs="markdown",
title="Dog Breed Classifier",
description="Upload an image, and the system will predict the breed and provide detailed information from the database.",
examples=['/content/drive/Othercomputers/我的 MacBook Pro/Learning/Cats_Dogs_Detector/Border_Collie.jpg',
'/content/drive/Othercomputers/我的 MacBook Pro/Learning/Cats_Dogs_Detector/Golden_Retriever.jpeg',
'/content/drive/Othercomputers/我的 MacBook Pro/Learning/Cats_Dogs_Detector/Saint_Bernard.jpeg',
'/content/drive/Othercomputers/我的 MacBook Pro/Learning/Cats_Dogs_Detector/French_Bulldog.jpeg',
'/content/drive/Othercomputers/我的 MacBook Pro/Learning/Cats_Dogs_Detector/Samoyed.jpg'],
css="""
.output-markdown {
font-family: Noto Sans, sans-serif;
line-height: 1.6;
}
.gr-button {
background-color: #3498DB;
color: white;
border-radius: 8px;
box-shadow: 0px 2px 4px rgba(0, 0, 0, 0.2);
padding: 10px 20px;
}
.gr-button:hover {
background-color: #2980B9;
}
.gr-box {
background: linear-gradient(to bottom, #f2f4f5, #ffffff);
border-radius: 10px;
padding: 20px;
box-shadow: 0px 4px 6px rgba(0, 0, 0, 0.1);
}
""",
theme="default"
)
# Launch the app
if __name__ == "__main__":
iface.launch()