app.py

gradio
import spaces
import gradio as gr
import torch
import numpy as np
from PIL import Image
from transformers import Sam3Processor, Sam3Model
import requests
import warnings
warnings.filterwarnings("ignore")

# Global model and processor
device = "cuda" if torch.cuda.is_available() else "cpu"
model = Sam3Model.from_pretrained("facebook/sam3", torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32).to(device)
processor = Sam3Processor.from_pretrained("facebook/sam3")

@spaces.GPU()
def segment(image: Image.Image, text: str, threshold: float, mask_threshold: float):
    """
    Perform promptable concept segmentation using SAM3.
    Returns format compatible with gr.AnnotatedImage: (image, [(mask, label), ...])
    """
    if image is None:
        return None, "📷 Please upload an image to begin."
    
    if not text.strip():
        return (image, []), "✏️ Enter a text prompt (e.g., 'person', 'cat', 'car')."
    
    try:
        inputs = processor(images=image, text=text.strip(), return_tensors="pt").to(device)
        
        for key in inputs:
            if inputs[key].dtype == torch.float32:
                inputs[key] = inputs[key].to(model.dtype)
        
        with torch.no_grad():
            outputs = model(**inputs)
        
        results = processor.post_process_instance_segmentation(
            outputs,
            threshold=threshold,
            mask_threshold=mask_threshold,
            target_sizes=inputs.get("original_sizes").tolist()
        )[0]
        
        n_masks = len(results['masks'])
        if n_masks == 0:
            return (image, []), f"🔍 No objects found for **'{text}'** — try adjusting thresholds."
        
        annotations = []
        for i, (mask, score) in enumerate(zip(results['masks'], results['scores'])):
            mask_np = mask.cpu().numpy().astype(np.float32)
            label = f"#{i+1} ({score:.2f})"
            annotations.append((mask_np, label))
        
        scores_text = ", ".join([f"{s:.2f}" for s in results['scores'].cpu().numpy()[:5]])
        info = f"✨ **{n_masks}** objects found for **'{text}'**\n\nConfidence: {scores_text}{'...' if n_masks > 5 else ''}"
        
        return (image, annotations), info
        
    except Exception as e:
        return (image, []), f"❌ Error: {str(e)}"

def clear_all():
    """Clear all inputs and outputs"""
    return None, "", None, 0.5, 0.5, "💡 Enter a prompt and click **Segment** to find objects."

def segment_example(image_path: str, prompt: str):
    """Handle example clicks"""
    if image_path.startswith("http"):
        image = Image.open(requests.get(image_path, stream=True).raw).convert("RGB")
    else:
        image = Image.open(image_path).convert("RGB")
    return segment(image, prompt, 0.5, 0.5)

# Custom modern theme
custom_theme = gr.themes.Glass(
    primary_hue="slate",
    secondary_hue="zinc",
    neutral_hue="slate",
    font=gr.themes.GoogleFont("Inter"),
    text_size="md",
    spacing_size="lg",
    radius_size="md"
).set(
    button_primary_background_fill="*neutral_800",
    button_primary_background_fill_hover="*neutral_700",
    button_secondary_background_fill="*neutral_100",
    button_secondary_background_fill_hover="*neutral_200",
    block_background_fill="white",
    block_secondary_background_fill="*neutral_50",
    block_title_text_weight="600",
)

# Main application
with gr.Blocks() as demo:
    # Header
    gr.HTML("""
    <div style="text-align: center; padding: 0.5rem 0; margin-bottom: 0.5rem;">
        <h1 style="font-size: 1.75rem; font-weight: 700; margin: 0; color: var(--neutral-800);">
            SAM3 <span style="font-weight: 400; color: var(--neutral-500);">Promptable Segmentation</span>
        </h1>
        <p style="margin: 0.25rem 0 0 0; color: var(--neutral-600); font-size: 0.875rem;">
            Zero-shot instance segmentation with natural language
        </p>
        <a href="https://huggingface.co/spaces/akhaliq/anycoder" 
           style="color: var(--primary-600); text-decoration: none; font-size: 0.8rem;"
           target="_blank">Built with anycoder</a>
    </div>
    """)
    
    # Main content
    with gr.Column(elem_classes=["main-content"]):
        # Image section
        with gr.Row(equal_height=True):
            with gr.Column(scale=1, min_width=280):
                image_input = gr.Image(
                    label="📷 Upload Image",
                    type="pil",
                    height=320,
                    sources=["upload", "clipboard"],
                )
            
            with gr.Column(scale=1, min_width=280):
                image_output = gr.AnnotatedImage(
                    label="🎯 Segmentation Result",
                    height=320,
                    show_legend=True,
                )
        
        # Info output
        info_output = gr.Markdown(
            value="💡 **Upload an image** and enter a prompt like 'person', 'cat', or 'car'",
            elem_classes=["info-box"],
        )
        
        # Controls section
        with gr.Group(elem_classes=["controls"]):
            with gr.Row():
                text_input = gr.Textbox(
                    label="What to find",
                    placeholder="e.g., person, cat, bicycle...",
                    scale=4,
                )
                segment_btn = gr.Button(
                    "🔍 Segment",
                    variant="primary",
                    size="lg",
                    scale=1,
                    min_width=120,
                )
            
            with gr.Row():
                thresh_slider = gr.Slider(
                    minimum=0.0,
                    maximum=1.0,
                    value=0.5,
                    step=0.01,
                    label="Detection",
                    info="Confidence threshold",
                    scale=1,
                )
                mask_thresh_slider = gr.Slider(
                    minimum=0.0,
                    maximum=1.0,
                    value=0.5,
                    step=0.01,
                    label="Mask",
                    info="Edge sharpness",
                    scale=1,
                )
                clear_btn = gr.Button(
                    "↺ Clear",
                    variant="secondary",
                    size="lg",
                    scale=0,
                    min_width=80,
                )
        
        # Examples
        gr.Markdown("### Quick Examples")
        gr.Examples(
            examples=[
                ["http://images.cocodataset.org/val2017/000000077595.jpg", "cat"],
                ["https://images.unsplash.com/photo-1535930483905-2c6d14342d7a", "dog"],
                ["https://images.unsplash.com/photo-1558618666-fcd25c85cd64", "car"],
            ],
            inputs=[image_input, text_input],
            outputs=[image_output, info_output],
            fn=segment_example,
            cache_examples=False,
            examples_per_page=3,
        )
        
        # Footer info
        gr.Accordion("ℹ️ About", open=False):
            gr.Markdown("""
            **SAM3** uses natural language prompts for zero-shot instance segmentation.
            
            - **Model**: [facebook/sam3](https://huggingface.co/facebook/sam3)
            - GPU recommended for faster processing
            - Works best with specific, clear object names
            """)

    # Event handlers
    segment_btn.click(
        fn=segment,
        inputs=[image_input, text_input, thresh_slider, mask_thresh_slider],
        outputs=[image_output, info_output],
        api_visibility="public",
    )
    
    clear_btn.click(
        fn=clear_all,
        outputs=[image_input, text_input, image_output, thresh_slider, mask_thresh_slider, info_output],
        api_visibility="private",
    )

# Custom CSS for responsive design
custom_css = """
@media (max-width: 768px) {
    .main-content {
        gap: 0.75rem !important;
    }
    .controls {
        gap: 0.75rem !important;
    }
    .info-box {
        font-size: 0.875rem !important;
        padding: 0.75rem !important;
    }
}

@media (max-width: 480px) {
    .gradio-group {
        gap: 0.5rem !important;
    }
}

.info-box {
    background: var(--neutral-50);
    border-radius: var(--radius-lg);
    padding: 1rem;
    border: 1px solid var(--neutral-200);
}

.controls {
    background: var(--neutral-50);
    border-radius: var(--radius-lg);
    padding: 1.25rem;
    border: 1px solid var(--neutral-200);
}

.gradio-annotatedimage {
    border: 2px dashed var(--neutral-300);
    border-radius: var(--radius-lg);
}

.gradio-group {
    gap: 1rem !important;
}
"""

if __name__ == "__main__":
    demo.launch(
        server_name="0.0.0.0",
        server_port=7860,
        share=False,
        debug=True,
        theme=custom_theme,
        css=custom_css,
        footer_links=[
            {"label": "anycoder", "url": "https://huggingface.co/spaces/akhaliq/anycoder"},
            {"label": "Model", "url": "https://huggingface.co/facebook/sam3"},
        ],
    )