Add initial A360 Croptool Gradio app scaffold

app.py

@@ -0,0 +1,163 @@
+import gradio as gr
+from PIL import Image
+from typing import Tuple, Optional
+
+# -------------------------------------------------------------------
+# A360 Croptool – Model-Agnostic Cropping Test Harness
+#
+# NOTE:
+# - This version only stubs the model calls.
+# - I will later plug in:
+#     • RetinaFace / InsightFace (face + landmarks)
+#     • YOLOv8/YOLOv9 (face/body/region detection)
+#     • Face Alignment nets
+#     • SAM / SAM-HQ
+#     • BiSeNet (face parsing)
+#     • CLIPSeg (text-guided cropping)
+# -------------------------------------------------------------------
+
+FACE_MODEL_CHOICES = [
+    "RetinaFace (InsightFace)",
+    "InsightFace 2D Landmarks (2d106)",
+    "Face-Alignment (1adrianb)",
+]
+
+BODY_MODEL_CHOICES = [
+    "YOLOv8 Body Detector",
+    "YOLOv9 Body / Part Detector",
+    "Human Pose (OpenPose-style)",
+]
+
+SEGMENTATION_MODEL_CHOICES = [
+    "SAM / SAM-HQ",
+    "BiSeNet Face Parsing",
+    "CLIPSeg (text-guided)",
+]
+
+CROP_TARGET_CHOICES = [
+    "Full Face",
+    "Eyes / Upper Face",
+    "Lips / Lower Face",
+    "Jawline / Chin",
+    "Neck",
+    "Chest / Breasts",
+    "Abdomen",
+    "Waist / Hips",
+    "Arms",
+    "Thighs / Legs",
+    "Custom Region",
+]
+
+
+def stub_crop(
+    image: Optional[Image.Image],
+    crop_target: str,
+    face_model: str,
+    body_model: str,
+    seg_model: str,
+    text_prompt: str,
+) -> Tuple[Image.Image, str]:
+    """
+    Placeholder cropping callback.
+
+    For now, simply returns the original image and a text summary of
+    the options the user selected. I'll replace this with real model
+    calls (RetinaFace / YOLO / SAM / CLIPSeg) later.
+    """
+    if image is None:
+        # Gradio will show this as an error toast
+        raise gr.Error("Please upload an image first.")
+
+    summary_lines = [
+        "Cropping request received:",
+        f"• Target region: {crop_target}",
+        f"• Face model: {face_model or 'None selected'}",
+        f"• Body model: {body_model or 'None selected'}",
+        f"• Segmentation model: {seg_model or 'None selected'}",
+    ]
+    if text_prompt.strip():
+        summary_lines.append(f"• Text prompt (for CLIPSeg / SAM): {text_prompt.strip()}")
+
+    summary_lines.append("")
+    summary_lines.append("NOTE: This is a stub implementation. "
+                          "Model hooks are ready; image is returned unmodified for now.")
+
+    return image, "\n".join(summary_lines)
+
+
+def create_app() -> gr.Blocks:
+    with gr.Blocks(theme="gradio/soft", css="""
+        .a360-header { font-size: 1.8rem; font-weight: 700; }
+        .a360-subtitle { opacity: 0.8; }
+    """) as demo:
+        gr.Markdown(
+            "<div class='a360-header'>A360 Croptool 🧬</div>"
+            "<div class='a360-subtitle'>"
+            "Test and prototype clinical cropping models for faces, bodies, and regions of interest."
+            "</div>"
+        )
+
+        with gr.Row():
+            with gr.Column(scale=1):
+                input_image = gr.Image(
+                    label="Input Image",
+                    type="pil",
+                    height=480,
+                    elem_id="input_image",
+                )
+
+                crop_target = gr.Dropdown(
+                    CROP_TARGET_CHOICES,
+                    value="Full Face",
+                    label="Target Region",
+                    info="What do you want to crop to?",
+                )
+
+                gr.Markdown("### Model Stack (configuration only – models wired later)")
+
+                face_model = gr.Radio(
+                    FACE_MODEL_CHOICES,
+                    value="RetinaFace (InsightFace)",
+                    label="Face / Landmark Model",
+                )
+
+                body_model = gr.Radio(
+                    BODY_MODEL_CHOICES,
+                    value="YOLOv8 Body Detector",
+                    label="Body / Region Model",
+                )
+
+                seg_model = gr.Radio(
+                    SEGMENTATION_MODEL_CHOICES,
+                    value="SAM / SAM-HQ",
+                    label="Segmentation / Mask Model",
+                )
+
+                text_prompt = gr.Textbox(
+                    label="Optional Text Prompt (for CLIPSeg / SAM)",
+                    placeholder="e.g., 'crop to lips', 'crop to abdomen', 'crop to jawline'",
+                    lines=2,
+                )
+
+                run_btn = gr.Button("Run Cropping Prototype", variant="primary")
+
+            with gr.Column(scale=1):
+                output_image = gr.Image(
+                    label="Cropped Output (stub – currently same as input)",
+                    height=480,
+                )
+                debug_text = gr.Markdown(label="Cropping Summary")
+
+        run_btn.click(
+            stub_crop,
+            inputs=[input_image, crop_target, face_model, body_model, seg_model, text_prompt],
+            outputs=[output_image, debug_text],
+        )
+
+        return demo
+
+
+app = create_app()
+
+if __name__ == "__main__":
+    app.launch()

requirements.txt

@@ -0,0 +1,4 @@
+gradio>=4.0.0
+Pillow
+numpy
+opencv-python