app.py

import os
import torch
import numpy as np
import cv2
import requests
import gradio as gr
import time 
from segment_anything import sam_model_registry, SamPredictor
import supervision as sv

# ------------------------------
# 1. Setup & Model Loading
# ------------------------------
device = "cuda" if torch.cuda.is_available() else "cpu"
print(f"Device used: {device}")

model_type = "vit_b"
checkpoint_path = "sam_vit_b_01ec64.pth"

# Download model if needed
if not os.path.exists(checkpoint_path):
    print("Downloading SAM checkpoint...")
    url = "https://dl.fbaipublicfiles.com/segment_anything/sam_vit_b_01ec64.pth"
    r = requests.get(url, stream=True)
    with open(checkpoint_path, "wb") as f:
        for chunk in r.iter_content(1024 * 1024):
            if chunk:
                f.write(chunk)

sam = sam_model_registry[model_type](checkpoint=checkpoint_path)
sam.to(device)

# Note: We kept the model in float32. 
# Using sam.half() often causes runtime type mismatch errors with SamPredictor 
# unless the input image is explicitly cast to half-precision manually.
predictor = SamPredictor(sam)

# ------------------------------
# 2. The API Function with Progress
# ------------------------------
# We add `progress=gr.Progress()` as a default argument. 
# Gradio automatically handles this when passed to the interface.
def run_sam_api(image_url, box_coords, progress=gr.Progress()):
    
    # --- LOGGING HELPER ---
    def log_step(step_num, total_steps, message):
        timestamp = time.strftime("%H:%M:%S")
        print(f"[{timestamp}] Step {step_num}/{total_steps}: {message}")
        # Update Gradio UI bar (0.0 to 1.0)
        progress(step_num / total_steps, desc=message)
    # ----------------------

    log_step(1, 5, f"Starting request for {image_url}")
    print(f"Received Box: {box_coords}")

    # 1. Download the Image
    try:
        headers = {'User-Agent': 'Mozilla/5.0'}
        resp = requests.get(image_url, stream=True, headers=headers).raw
        image_array = np.asarray(bytearray(resp.read()), dtype="uint8")
        image = cv2.imdecode(image_array, cv2.IMREAD_COLOR)
        
        if image is None:
            raise ValueError("Could not decode image.")
            
        image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
    except Exception as e:
        print(f"Error downloading image: {e}")
        raise gr.Error(f"Failed to load image: {e}")

    # 2. Set Image for SAM (The Heavy Step)
    log_step(2, 5, "Generating Image Embeddings (This may take a moment)...")
    
    # Note: This specific line is where the GPU works hardest. 
    # It might 'freeze' here for a few seconds.
    predictor.set_image(image)

    # 3. Prepare Box
    input_box = np.array(box_coords)

    # 4. Predict
    log_step(3, 5, "Decoding Masks...")
    masks, scores, logits = predictor.predict(
        point_coords=None,
        point_labels=None,
        box=input_box[None, :], 
        multimask_output=False
    )

    # 5. Annotate / Visualize
    log_step(4, 5, "Annotating Image...")
    
    detections = sv.Detections(
        xyxy=sv.mask_to_xyxy(masks=masks),
        mask=masks
    )
    # Fix for missing class_id
    detections.class_id = np.zeros(len(detections), dtype=int)

    mask_annotator = sv.MaskAnnotator(color=sv.Color.RED)
    box_annotator = sv.BoxAnnotator(color=sv.Color.RED)

    annotated_image = mask_annotator.annotate(scene=image.copy(), detections=detections)
    annotated_image = box_annotator.annotate(scene=annotated_image, detections=detections)
    
    log_step(5, 5, "Done!")
    return annotated_image

demo = gr.Interface(
    fn=run_sam_api,
    inputs=[
        gr.Textbox(label="Image URL", placeholder="http://..."), 
        gr.JSON(label="Box Coords [x1, y1, x2, y2]", value=[2434, 666, 3737, 1756])
    ],
    outputs=gr.Image(type="numpy", label="Segmented Output"),
    title="SAM API via Gradio",
    description="Send an image URL and bounding box coordinates to segment objects.",
    api_name="predict_api"
)

if __name__ == "__main__":
    # If running on Hugging Face Spaces, just use launch()
    demo.queue().launch()