Spaces:
Sleeping
Sleeping
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,128 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import os
|
2 |
+
import torch
|
3 |
+
from torch.utils.data import Dataset, DataLoader
|
4 |
+
from PIL import Image
|
5 |
+
import torchvision
|
6 |
+
from torchvision import transforms
|
7 |
+
import xml.etree.ElementTree as ET
|
8 |
+
import torch.optim as optim
|
9 |
+
import matplotlib.pyplot as plt
|
10 |
+
import gradio as gr
|
11 |
+
|
12 |
+
# Ensure device is set to GPU if available
|
13 |
+
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
14 |
+
|
15 |
+
class FaceMaskDataset(Dataset):
|
16 |
+
def __init__(self, images_dir, annotations_dir, transform=None, resize=(800, 800)):
|
17 |
+
self.images_dir = images_dir
|
18 |
+
self.annotations_dir = annotations_dir
|
19 |
+
self.transform = transform
|
20 |
+
self.resize = resize
|
21 |
+
self.image_files = os.listdir(images_dir)
|
22 |
+
|
23 |
+
def __len__(self):
|
24 |
+
return len(self.image_files)
|
25 |
+
|
26 |
+
def __getitem__(self, idx):
|
27 |
+
image_path = os.path.join(self.images_dir, self.image_files[idx])
|
28 |
+
image = Image.open(image_path).convert("RGB")
|
29 |
+
|
30 |
+
# Resize the image to a fixed size, while maintaining aspect ratio
|
31 |
+
image = image.resize(self.resize)
|
32 |
+
|
33 |
+
# Handle both .jpg and .png files
|
34 |
+
annotation_path = os.path.join(self.annotations_dir, self.image_files[idx].replace(".jpg", ".xml").replace(".png", ".xml"))
|
35 |
+
|
36 |
+
if not os.path.exists(annotation_path):
|
37 |
+
print(f"Warning: Annotation file {annotation_path} does not exist. Skipping image {self.image_files[idx]}.")
|
38 |
+
return None, None # Return a tuple with None to skip the image/annotation pair
|
39 |
+
|
40 |
+
boxes, labels = self.load_annotations(annotation_path)
|
41 |
+
|
42 |
+
if boxes is None or labels is None:
|
43 |
+
return None, None # Skip this item if annotations are invalid
|
44 |
+
|
45 |
+
target = {'boxes': boxes, 'labels': labels}
|
46 |
+
|
47 |
+
if self.transform:
|
48 |
+
image = self.transform(image)
|
49 |
+
|
50 |
+
return image, target
|
51 |
+
|
52 |
+
def load_annotations(self, annotation_path):
|
53 |
+
tree = ET.parse(annotation_path)
|
54 |
+
root = tree.getroot()
|
55 |
+
|
56 |
+
boxes = []
|
57 |
+
labels = []
|
58 |
+
for obj in root.iter('object'):
|
59 |
+
label = obj.find('name').text
|
60 |
+
bndbox = obj.find('bndbox')
|
61 |
+
xmin = float(bndbox.find('xmin').text)
|
62 |
+
ymin = float(bndbox.find('ymin').text)
|
63 |
+
xmax = float(bndbox.find('xmax').text)
|
64 |
+
ymax = float(bndbox.find('ymax').text)
|
65 |
+
boxes.append([xmin, ymin, xmax, ymax])
|
66 |
+
labels.append(1 if label == "mask" else 0) # Assuming "mask" = 1, "no_mask" = 0
|
67 |
+
|
68 |
+
if len(boxes) == 0 or len(labels) == 0:
|
69 |
+
return None, None # If no boxes or labels are found, return None
|
70 |
+
|
71 |
+
boxes = torch.as_tensor(boxes, dtype=torch.float32)
|
72 |
+
labels = torch.tensor(labels, dtype=torch.int64)
|
73 |
+
|
74 |
+
return boxes, labels
|
75 |
+
|
76 |
+
|
77 |
+
# Define the collate function for DataLoader
|
78 |
+
def collate_fn(batch):
|
79 |
+
# Filter out None values and pack the rest into a batch
|
80 |
+
batch = [item for item in batch if item[0] is not None and item[1] is not None]
|
81 |
+
return tuple(zip(*batch))
|
82 |
+
|
83 |
+
# Load your pre-trained model (or initialize if required)
|
84 |
+
def load_model():
|
85 |
+
model = torchvision.models.detection.fasterrcnn_resnet50_fpn(pretrained=True)
|
86 |
+
# Assuming 2 classes: mask and no_mask
|
87 |
+
num_classes = 2
|
88 |
+
in_features = model.roi_heads.box_predictor.cls_score.in_features
|
89 |
+
model.roi_heads.box_predictor = torchvision.models.detection.faster_rcnn.FastRCNNPredictor(in_features, num_classes)
|
90 |
+
model.to(device)
|
91 |
+
return model
|
92 |
+
|
93 |
+
# Inference function
|
94 |
+
def infer(image):
|
95 |
+
model = load_model() # Load the model
|
96 |
+
model.eval()
|
97 |
+
|
98 |
+
# Apply transformations
|
99 |
+
transform = transforms.Compose([
|
100 |
+
transforms.Resize((224, 224)), # Resize all images to 224x224
|
101 |
+
transforms.ToTensor(),
|
102 |
+
])
|
103 |
+
|
104 |
+
image = Image.fromarray(image)
|
105 |
+
image = transform(image).unsqueeze(0).to(device) # Add batch dimension
|
106 |
+
|
107 |
+
with torch.no_grad():
|
108 |
+
prediction = model(image)
|
109 |
+
|
110 |
+
# Get boxes and labels from the predictions
|
111 |
+
boxes = prediction[0]['boxes'].cpu().numpy()
|
112 |
+
labels = prediction[0]['labels'].cpu().numpy()
|
113 |
+
|
114 |
+
return boxes, labels
|
115 |
+
|
116 |
+
# Gradio interface
|
117 |
+
def gradio_interface(image):
|
118 |
+
boxes, labels = infer(image)
|
119 |
+
|
120 |
+
# Assuming labels: 0 = no mask, 1 = mask
|
121 |
+
result = {"boxes": boxes, "labels": labels}
|
122 |
+
return result
|
123 |
+
|
124 |
+
# Create Gradio interface
|
125 |
+
iface = gr.Interface(fn=gradio_interface, inputs=gr.Image(type="numpy"), outputs="json")
|
126 |
+
|
127 |
+
# Launch Gradio interface
|
128 |
+
iface.launch()
|