Upload 3 files

app.py

@@ -0,0 +1,86 @@
+import torch
+import torch.nn as nn
+import torchvision.models as models
+import torchvision.transforms as T
+import gradio as gr
+import numpy as np
+import cv2
+from PIL import Image
+from sklearn.metrics.pairwise import cosine_similarity
+
+# === Model ===
+class SimpleUNet(nn.Module):
+    def __init__(self):
+        super(SimpleUNet, self).__init__()
+        base_model = models.mobilenet_v2(pretrained=True).features
+        self.encoder = base_model
+        self.decoder = nn.Sequential(
+            nn.ConvTranspose2d(1280, 512, kernel_size=2, stride=2),
+            nn.ReLU(),
+            nn.ConvTranspose2d(512, 256, kernel_size=2, stride=2),
+            nn.ReLU(),
+            nn.ConvTranspose2d(256, 128, kernel_size=2, stride=2),
+            nn.ReLU(),
+            nn.ConvTranspose2d(128, 64, kernel_size=2, stride=2),
+            nn.ReLU(),
+            nn.ConvTranspose2d(64, 1, kernel_size=2, stride=2),
+            nn.Sigmoid()
+        )
+
+    def forward(self, x):
+        x = self.encoder(x)
+        x = self.decoder(x)
+        return x
+
+# === Load Model ===
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+model = SimpleUNet().to(device)
+model.load_state_dict(torch.load("iris_segmentation_model.pth", map_location=device))
+model.eval()
+
+# === Transform ===
+transform = T.Compose([
+    T.Resize((224, 224)),
+    T.ToTensor(),
+    T.Normalize([0.5, 0.5, 0.5], [0.5, 0.5, 0.5])
+])
+
+# === Main function ===
+def segment_iris(image):
+    frame = np.array(image)
+    gray = cv2.cvtColor(frame, cv2.COLOR_RGB2GRAY)
+    clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8, 8))
+    enhanced = clahe.apply(gray)
+    rgb_enhanced = cv2.cvtColor(enhanced, cv2.COLOR_GRAY2RGB)
+    pil_img = Image.fromarray(rgb_enhanced)
+
+    input_tensor = transform(pil_img).unsqueeze(0).to(device)
+
+    with torch.no_grad():
+        pred = model(input_tensor)
+        pred_mask = pred.squeeze().cpu().numpy()
+
+    binary_mask = (pred_mask > 0.5).astype(np.uint8) * 255
+    binary_mask = cv2.resize(binary_mask, (frame.shape[1], frame.shape[0]))
+    color_mask = cv2.applyColorMap(binary_mask, cv2.COLORMAP_JET)
+    blended = cv2.addWeighted(frame, 0.7, color_mask, 0.3, 0)
+
+    return Image.fromarray(blended)
+
+# === Gradio UI ===
+with gr.Blocks(theme=gr.themes.Soft()) as demo:
+    gr.Markdown("# 👁️ Iris Segmentation App")
+    gr.Markdown("Upload an eye image below. The model will perform iris segmentation and return a blended visualization. Webcam option only works when deployed locally.")
+
+    with gr.Row():
+        img_input = gr.Image(type="pil", label="📤 Upload Eye Image")
+        img_output = gr.Image(label="🎯 Segmentation Output")
+
+    img_input.change(fn=segment_iris, inputs=img_input, outputs=img_output)
+
+    gr.Markdown("⚠️ Webcam support is available only in local mode.")
+
+    # Placeholder for webcam mode
+    gr.Image(source="webcam", streaming=True, label="(Webcam – Local Only)", visible=False)
+
+demo.launch()

iris_segmentation_model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:769ed1ea4348ed4c273f477557fc46355575e1b6a92f88c40e45fff8c3600a41
+size 22390414

requirements.txt

@@ -0,0 +1,6 @@
+torch
+torchvision
+gradio
+opencv-python
+scikit-learn
+pillow