Delete app.py

app.py

@@ -1,263 +0,0 @@
-import time
-from typing import List, Tuple
-
-import cv2
-import gradio as gr
-import numpy as np
-
-# ------------------------------------------------------------
-# Phase 1: live webcam -> face bounding boxes -> beautiful UI
-# The emotion / age / demographic inference modules will be added later.
-# ------------------------------------------------------------
-
-CASCADE_PATH = cv2.data.haarcascades + "haarcascade_frontalface_default.xml"
-FACE_CASCADE = cv2.CascadeClassifier(CASCADE_PATH)
-
-APP_TITLE = "FaceSense Live"
-APP_SUBTITLE = "Real-time face box prototype for the Hugging Face Spaces deployment loop"
-
-
-def _draw_corner_box(
-    image: np.ndarray,
-    x: int,
-    y: int,
-    w: int,
-    h: int,
-    color: Tuple[int, int, int] = (52, 235, 186),
-    thickness: int = 3,
-    corner_ratio: float = 0.28,
-) -> None:
-    """Draw a modern corner-style bounding box directly on an RGB image."""
-    corner_w = max(18, int(w * corner_ratio))
-    corner_h = max(18, int(h * corner_ratio))
-
-    # Top-left
-    cv2.line(image, (x, y), (x + corner_w, y), color, thickness, cv2.LINE_AA)
-    cv2.line(image, (x, y), (x, y + corner_h), color, thickness, cv2.LINE_AA)
-
-    # Top-right
-    cv2.line(image, (x + w, y), (x + w - corner_w, y), color, thickness, cv2.LINE_AA)
-    cv2.line(image, (x + w, y), (x + w, y + corner_h), color, thickness, cv2.LINE_AA)
-
-    # Bottom-left
-    cv2.line(image, (x, y + h), (x + corner_w, y + h), color, thickness, cv2.LINE_AA)
-    cv2.line(image, (x, y + h), (x, y + h - corner_h), color, thickness, cv2.LINE_AA)
-
-    # Bottom-right
-    cv2.line(image, (x + w, y + h), (x + w - corner_w, y + h), color, thickness, cv2.LINE_AA)
-    cv2.line(image, (x + w, y + h), (x + w, y + h - corner_h), color, thickness, cv2.LINE_AA)
-
-
-def _draw_label(image: np.ndarray, x: int, y: int, lines: List[str]) -> None:
-    """Draw a translucent label card above a face box."""
-    font = cv2.FONT_HERSHEY_SIMPLEX
-    font_scale = 0.55
-    thickness = 1
-    padding_x = 10
-    padding_y = 8
-    line_height = 21
-
-    widths = [cv2.getTextSize(line, font, font_scale, thickness)[0][0] for line in lines]
-    card_w = max(widths) + padding_x * 2
-    card_h = line_height * len(lines) + padding_y * 2
-
-    y1 = max(8, y - card_h - 10)
-    x1 = max(8, min(x, image.shape[1] - card_w - 8))
-    x2 = min(image.shape[1] - 8, x1 + card_w)
-    y2 = min(image.shape[0] - 8, y1 + card_h)
-
-    overlay = image.copy()
-    cv2.rectangle(overlay, (x1, y1), (x2, y2), (14, 18, 31), -1)
-    cv2.addWeighted(overlay, 0.72, image, 0.28, 0, image)
-    cv2.rectangle(image, (x1, y1), (x2, y2), (52, 235, 186), 1, cv2.LINE_AA)
-
-    for i, line in enumerate(lines):
-        text_y = y1 + padding_y + 16 + i * line_height
-        cv2.putText(image, line, (x1 + padding_x, text_y), font, font_scale, (242, 250, 255), thickness, cv2.LINE_AA)
-
-
-def _detect_faces(frame_rgb: np.ndarray) -> List[Tuple[int, int, int, int]]:
-    """Detect frontal faces with OpenCV's built-in Haar cascade."""
-    gray = cv2.cvtColor(frame_rgb, cv2.COLOR_RGB2GRAY)
-    gray = cv2.equalizeHist(gray)
-
-    faces = FACE_CASCADE.detectMultiScale(
-        gray,
-        scaleFactor=1.08,
-        minNeighbors=5,
-        minSize=(70, 70),
-        flags=cv2.CASCADE_SCALE_IMAGE,
-    )
-
-    # Largest faces first makes the UI feel stable when several faces exist.
-    faces_list = [(int(x), int(y), int(w), int(h)) for x, y, w, h in faces]
-    return sorted(faces_list, key=lambda b: b[2] * b[3], reverse=True)
-
-
-def analyze_frame(frame: np.ndarray):
-    """Process one webcam frame. Gradio calls this repeatedly while streaming."""
-    if frame is None:
-        return None, "### Waiting for camera\nClick **Record from webcam** to start Phase 1."
-
-    start = time.perf_counter()
-    output = frame.copy()
-    faces = _detect_faces(output)
-
-    for idx, (x, y, w, h) in enumerate(faces, start=1):
-        _draw_corner_box(output, x, y, w, h)
-        _draw_label(
-            output,
-            x,
-            y,
-            [
-                f"Face {idx}",
-                "Expression: Phase 2",
-                "Age range: Phase 3",
-            ],
-        )
-
-    elapsed_ms = (time.perf_counter() - start) * 1000
-    fps = 1000 / elapsed_ms if elapsed_ms > 0 else 0
-
-    if faces:
-        status = f"""
-### Live status
-
-<div class='metric-grid'>
-  <div class='metric-card'><span class='metric-number'>{len(faces)}</span><span class='metric-label'>face(s) detected</span></div>
-  <div class='metric-card'><span class='metric-number'>{elapsed_ms:.1f} ms</span><span class='metric-label'>processing time</span></div>
-  <div class='metric-card'><span class='metric-number'>{fps:.1f}</span><span class='metric-label'>estimated FPS</span></div>
-</div>
-
-**Phase 1 result:** webcam streaming and bounding boxes are active.  
-**Next:** add expression model, smoothing, and calibrated confidence labels.
-"""
-    else:
-        status = f"""
-### Live status
-
-<div class='metric-grid'>
-  <div class='metric-card'><span class='metric-number'>0</span><span class='metric-label'>faces detected</span></div>
-  <div class='metric-card'><span class='metric-number'>{elapsed_ms:.1f} ms</span><span class='metric-label'>processing time</span></div>
-</div>
-
-Move closer to the camera, face the lens, and use decent lighting. Phase 1 uses a lightweight detector so we can confirm deployment first.
-"""
-
-    return output, status
-
-
-CUSTOM_CSS = """
-.gradio-container {
-    background:
-        radial-gradient(circle at top left, rgba(72, 219, 251, 0.22), transparent 34%),
-        radial-gradient(circle at bottom right, rgba(163, 255, 181, 0.16), transparent 34%),
-        linear-gradient(135deg, #070b16 0%, #111827 45%, #0b1020 100%) !important;
-    color: #f8fafc !important;
-}
-.hero {
-    text-align: center;
-    padding: 24px 20px 10px 20px;
-}
-.hero h1 {
-    font-size: 3rem;
-    line-height: 1.05;
-    margin-bottom: 8px;
-    letter-spacing: -0.04em;
-}
-.hero p {
-    color: #cbd5e1;
-    font-size: 1.05rem;
-    margin: 0 auto;
-    max-width: 760px;
-}
-.panel {
-    border: 1px solid rgba(148, 163, 184, 0.24) !important;
-    border-radius: 22px !important;
-    background: rgba(15, 23, 42, 0.74) !important;
-    box-shadow: 0 18px 70px rgba(0, 0, 0, 0.38) !important;
-    backdrop-filter: blur(18px) !important;
-    overflow: hidden !important;
-}
-.metric-grid {
-    display: grid;
-    grid-template-columns: repeat(auto-fit, minmax(120px, 1fr));
-    gap: 12px;
-    margin: 12px 0 16px;
-}
-.metric-card {
-    padding: 14px;
-    border-radius: 16px;
-    background: rgba(15, 23, 42, 0.78);
-    border: 1px solid rgba(72, 219, 251, 0.28);
-}
-.metric-number {
-    display: block;
-    font-size: 1.4rem;
-    font-weight: 800;
-    color: #7dd3fc;
-}
-.metric-label {
-    display: block;
-    color: #cbd5e1;
-    font-size: 0.82rem;
-    margin-top: 2px;
-}
-.footer-note {
-    text-align: center;
-    color: #94a3b8;
-    font-size: 0.92rem;
-    padding: 12px 0 4px;
-}
-"""
-
-with gr.Blocks(css=CUSTOM_CSS, theme=gr.themes.Soft(primary_hue="cyan", neutral_hue="slate")) as demo:
-    gr.HTML(
-        f"""
-        <div class='hero'>
-            <h1>{APP_TITLE}</h1>
-            <p>{APP_SUBTITLE}</p>
-        </div>
-        """
-    )
-
-    with gr.Row(equal_height=True):
-        with gr.Column(scale=1, elem_classes=["panel"]):
-            webcam = gr.Image(
-                label="Camera input",
-                sources=["webcam"],
-                type="numpy",
-                streaming=True,
-                height=520,
-            )
-        with gr.Column(scale=1, elem_classes=["panel"]):
-            annotated = gr.Image(
-                label="Annotated output",
-                type="numpy",
-                height=520,
-            )
-
-    with gr.Row():
-        with gr.Column(elem_classes=["panel"]):
-            status = gr.Markdown(
-                "### Waiting for camera\nClick **Record from webcam** to start Phase 1.",
-                elem_id="status-card",
-            )
-
-    gr.HTML(
-        "<div class='footer-note'>Phase 1 validates live webcam processing and deployment. Prediction models come next after this passes your test.</div>"
-    )
-
-    webcam.stream(
-        analyze_frame,
-        inputs=webcam,
-        outputs=[annotated, status],
-        time_limit=30,
-        stream_every=0.12,
-        concurrency_limit=10,
-        api_name="analyze_frame",
-    )
-
-if __name__ == "__main__":
-    demo.queue(default_concurrency_limit=10)
-    demo.launch()