feat: add complete Gradio pose detection app using MoveNet

app.py

@@ -0,0 +1,46 @@
+import gradio as gr
+import PIL.Image
+from poser import draw_bones, movenet
+import tensorflow as tf
+import numpy as np
+
+def predict(image: PIL.Image.Image):
+    input_size = 256
+    image = image.resize((1280, 1280))
+    image_tf = tf.keras.preprocessing.image.img_to_array(image)
+
+    # Prepare input for the model
+    input_image = tf.expand_dims(image_tf, axis=0)
+    input_image = tf.image.resize_with_pad(input_image, input_size, input_size)
+
+    # Run MoveNet pose estimation
+    keypoints = movenet(input_image)
+
+    # Draw bones on the image
+    joints = draw_bones(image, keypoints)
+
+    # Format points as text
+    points = [f"{label} → ({int(float(x))}, {int(float(y))})" for label, x, y in joints]
+
+    return image, joints, points
+
+with gr.Blocks(title="MoveNet Pose Estimation") as demo:
+    gr.Markdown("# 🧍‍♀️ Human Pose Estimation with MoveNet")
+    gr.Markdown("Upload an image to detect body keypoints and view the skeleton overlay.")
+
+    with gr.Row():
+        with gr.Column():
+            input_image = gr.Image(type="pil", label="Input Image")
+            run_button = gr.Button("Detect Pose", variant="primary")
+        with gr.Column():
+            output_image = gr.Image(type="numpy", label="Skeleton Output")
+            joints_table = gr.Dataframe(headers=["Label", "X", "Y"], row_count=17, col_count=(3, "fixed"))
+            point_text = gr.Textbox(label="Formatted Keypoints", lines=8)
+
+    run_button.click(
+        fn=predict,
+        inputs=[input_image],
+        outputs=[output_image, joints_table, point_text]
+    )
+
+demo.launch()

download_movenet.py

@@ -1,9 +0,0 @@
-import tensorflow as tf
-import tensorflow_hub as hub
-import os
-
-save_path = "movenet_saved_model"
-model_url = "https://tfhub.dev/google/movenet/singlepose/thunder/4"
-model = hub.load(model_url)
-tf.saved_model.save(model, export_dir=save_path)
-print(f"Model saved to: {os.path.abspath(save_path)}")

poser.py

@@ -0,0 +1,67 @@
+import tensorflow as tf
+import numpy as np
+from PIL import ImageDraw, ImageFont
+from huggingface_hub import snapshot_download
+
+# Keypoint labels for MoveNet (17 human body parts)
+KEYPOINT_LABELS = {
+    'nose': 0, 'left_eye': 1, 'right_eye': 2, 'left_ear': 3, 'right_ear': 4,
+    'left_shoulder': 5, 'right_shoulder': 6, 'left_elbow': 7, 'right_elbow': 8,
+    'left_wrist': 9, 'right_wrist': 10, 'left_hip': 11, 'right_hip': 12,
+    'left_knee': 13, 'right_knee': 14, 'left_ankle': 15, 'right_ankle': 16
+}
+
+# Stylish connection colors for the skeleton bones
+SKELETON_EDGES = {
+    (0, 1): '#FF66CC', (0, 2): '#66FFFF', (1, 3): '#FF66CC', (2, 4): '#66FFFF',
+    (0, 5): '#FF99CC', (0, 6): '#99FFFF', (5, 7): '#FF6699', (7, 9): '#FF3366',
+    (6, 8): '#66CCCC', (8, 10): '#33CCCC', (5, 6): '#CCCC00', (5, 11): '#FF9966',
+    (6, 12): '#66FF99', (11, 12): '#999900', (11, 13): '#FF6600', (13, 15): '#FF3300',
+    (12, 14): '#00CC99', (14, 16): '#009966'
+}
+
+def process_keypoints(prediction, img_height, img_width, confidence=0.12):
+    all_joints = []
+    all_bones = []
+
+    instance_count = prediction.shape[1]
+    for i in range(instance_count):
+        x_coords = prediction[0, i, :, 1] * img_width
+        y_coords = prediction[0, i, :, 0] * img_height
+        scores = prediction[0, i, :, 2]
+        labels = list(KEYPOINT_LABELS.keys())
+
+        keypoints = np.stack([labels, x_coords, y_coords], axis=-1)
+        visible_kpts = keypoints[scores > confidence]
+        all_joints.append(visible_kpts)
+
+        for (a, b), color in SKELETON_EDGES.items():
+            if scores[a] > confidence and scores[b] > confidence:
+                segment = np.array([[x_coords[a], y_coords[a]], [x_coords[b], y_coords[b]]])
+                all_bones.append((segment, color))
+
+    return np.concatenate(all_joints, axis=0), all_bones
+
+def draw_bones(image, keypoints):
+    draw = ImageDraw.Draw(image)
+    font = ImageFont.load_default()
+
+    joints, bones = process_keypoints(keypoints, image.height, image.width)
+
+    for bone, color in bones:
+        draw.line((*bone[0], *bone[1]), fill=color, width=3)
+
+    for label, x, y in joints:
+        cx, cy = float(x), float(y)
+        radius = 4
+        draw.ellipse([(cx - radius, cy - radius), (cx + radius, cy + radius)], fill="#FF0000", outline="#222222")
+        draw.text((cx + 5, cy - 5), label, font=font, fill="#0000CC")
+
+    return joints
+
+def movenet(image):
+    model_path = snapshot_download("lukassso/movenet-myking")
+    model = tf.saved_model.load(model_path).signatures["serving_default"]
+    image = tf.cast(image, tf.int32)
+    result = model(image)
+    return result["output_0"].numpy()