Merge pull request #99 from intel/update-branch

292 user story include stable diffusion xl in text to image generation microservice (#293)

Author: gooishin

usecases/ai/microservices/text-to-image/README.md

@@ -5,6 +5,7 @@ Text-to-image generation microservice is a FastAPI-based API that interacts with
 It provides endpoints for managing the pipeline, checking the pipeline status, retrieving the generated image, and performing a health check.
 
 ### Supported Models
+* Stable Diffusion XL
 * Stable Diffusion v3.5 
 * Stable Diffusion v3
 

usecases/ai/microservices/text-to-image/stable-diffusion-v3.5/.gitignore

@@ -0,0 +1,3 @@
+.venv
+stable-diffusion-3.5-medium
+kernel.errors.txt

usecases/ai/microservices/text-to-image/stable-diffusion-v3.5/requirements.txt

@@ -1,10 +1,10 @@
 git+https://github.com/huggingface/optimum-intel.git
-openvino>=2024.6
-openvino-tokenizers>=2024.6
-openvino-genai>=2024.6
-Pillow
-diffusers>=0.30.3
-gradio>=4.19
-typing_extensions>=4.9
-numpy<2.0.0; platform_system == "Darwin"
-IPython
+openvino==2024.6.0
+openvino-tokenizers==2024.6.0.0
+openvino-genai==2024.6.0.0
+Pillow==11.1.0
+diffusers==0.32.1
+gradio==5.11.0
+typing_extensions==4.12.2
+numpy==2.1.3
+IPython==8.31.0

usecases/ai/microservices/text-to-image/stable-diffusion-v3/.gitignore

@@ -0,0 +1,3 @@
+.venv
+stable-diffusion-3
+kernel.errors.txt

usecases/ai/microservices/text-to-image/stable-diffusion-v3/backend/client.py

@@ -6,7 +6,7 @@
 
 
 class PipelineClient:
-    def __init__(self, base_url="http://localhost:8000"):
+    def __init__(self, base_url="http://localhost:8100"):
         """
         Initialize the PipelineClient with the base URL.
         :param base_url: The base URL for the API.

usecases/ai/microservices/text-to-image/stable-diffusion-v3/backend/server.py

@@ -318,4 +318,4 @@ def health_check():
 
 if __name__ == "__main__":
     api = Sdv3API()
-    uvicorn.run(api.app, host="0.0.0.0", port=8000, reload=False)
+    uvicorn.run(api.app, host="0.0.0.0", port=8100, reload=False)

usecases/ai/microservices/text-to-image/stable-diffusion-v3/requirements.txt

@@ -4,22 +4,22 @@
 git+https://github.com/initml/diffusers.git@clement/feature/flash_sd3
 
 # Dependencies with version specifications
-gradio>=4.19
-torch>=2.1
-transformers
-nncf>=2.12.0
-datasets>=2.14.6
-opencv-python
-pillow
-peft>=0.7.0
-ipywidgets
+gradio==5.11.0
+torch==2.5.1+cpu
+transformers==4.47.1
+nncf==2.14.1
+datasets==3.2.0
+opencv-python==4.10.0.84
+pillow==11.1.0
+peft==0.14.0
+ipywidgets==8.1.5
 
 # Additional index URL for PyTorch dependencies
 --extra-index-url https://download.pytorch.org/whl/cpu
 
 # OpenVINO with version specification
-openvino>=2024.3.0
+openvino==2024.6.0
 
 # Server
-fastapi
-uvicorn
+fastapi==0.115.6
+uvicorn==0.34.0

usecases/ai/microservices/text-to-image/stable-diffusion-xl/.gitignore

@@ -0,0 +1,3 @@
+.venv
+openvino-sd-xl-base-1.0
+kernel.errors.txt

usecases/ai/microservices/text-to-image/stable-diffusion-xl/backend/client.py

@@ -0,0 +1,126 @@
+# Copyright (C) 2025 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+
+import time
+import requests
+
+
+class PipelineClient:
+    def __init__(self, base_url="http://localhost:8100"):
+        """
+        Initialize the PipelineClient with the base URL.
+        :param base_url: The base URL for the API.
+        """
+        self.base_url = base_url
+
+    def make_request(self, method, endpoint, data=None, timeout=100):
+        """
+        Make a request to the API.
+        :param method: HTTP method (GET, POST, etc.).
+        :param endpoint: API endpoint.
+        :param data: JSON payload for the request.
+        :param timeout: Timeout in seconds.
+        :return: Response object or None on failure.
+        """
+        url = f"{self.base_url}{endpoint}"
+        try:
+            response = requests.request(method, url, json=data, timeout=timeout)
+            response.raise_for_status()
+            return response
+        except requests.RequestException as e:
+            print(f"Error while accessing {url}: {e}")
+            return None
+
+    @staticmethod
+    def save_image(content, filename):
+        """
+        Save the content as an image file.
+        :param content: Binary content of the image.
+        :param filename: Filename to save the image.
+        """
+        try:
+            with open(filename, "wb") as file:
+                file.write(content)
+            print(f"Image saved as {filename}")
+        except IOError as e:
+            print(f"Failed to save image: {e}")
+
+    def wait_for_completion(self, timeout=300, poll_interval=2):
+        """
+        Wait for the pipeline to complete.
+        :param timeout: Maximum time to wait in seconds.
+        :param poll_interval: Time between status checks.
+        :return: True if completed, False otherwise.
+        """
+        start_time = time.time()
+        while time.time() - start_time < timeout:
+            response = self.make_request("GET", "/pipeline/status", timeout=10)
+            if response:
+                status = response.json()
+                print(f"Pipeline status: Running={status['running']}, Completed={status['completed']}")
+                if status["completed"]:
+                    return True
+            time.sleep(poll_interval)
+        print("Timeout reached. Pipeline execution did not complete.")
+        return False
+
+    def check_health(self):
+        """
+        Check the health of the API.
+        :return: True if the health check passes, False otherwise.
+        """
+        response = self.make_request("GET", "/health")
+        if response:
+            health_status = response.json()
+            print(f"Health status: {health_status['status']}")
+            return health_status['status'] == "healthy"
+        else:
+            print("Failed to perform health check.")
+            return False
+
+
+
+def main():
+    client = PipelineClient()
+
+    # Step 1: Health check
+    if not client.check_health():
+        print("Health check failed. Exiting.")
+        return
+
+    # # Step 2: Select the device
+    # response = client.make_request("POST", "/pipeline/select-device", {"device": "CPU"})
+    # if response:
+    #     print(response.json())
+
+    # Step 3: Trigger the pipeline with additional parameters
+    response = client.make_request(
+        "POST",
+        "/pipeline/run",
+        {
+            "prompt": "A raccoon trapped inside a glass jar full of colorful candies, the background is steamy with vivid colors",
+            "width": 1024,  # Additional parameter: width
+            "height": 1024,  # Additional parameter: height
+            "num_inference_steps": 50  # Additional parameter: num_inference_steps
+        },
+    )
+    if not response:
+        print("Failed to trigger the pipeline.")
+        return
+    print(response.json())
+
+    # Step 4: Wait for completion
+    if not client.wait_for_completion():
+        return
+
+    # # Step 5: Retrieve the generated image
+    # response = client.make_request("GET", "/pipeline/image")
+    # if response:
+    #     client.save_image(response.content, "output_image.png")
+    # else:
+    #     print("Failed to retrieve the generated image.")
+
+
+if __name__ == "__main__":
+    main()
+

usecases/ai/microservices/text-to-image/stable-diffusion-xl/backend/cmd_helper.py

@@ -0,0 +1,43 @@
+# Copyright (C) 2025 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+
+import subprocess  # nosec - disable B404:import-subprocess check
+import sys
+from pathlib import Path
+from typing import Dict
+import platform
+
+
+def clone_repo(repo_url: str, revision: str = None, add_to_sys_path: bool = True) -> Path:
+    repo_path = Path(repo_url.split("/")[-1].replace(".git", ""))
+
+    if not repo_path.exists():
+        try:
+            subprocess.run(["git", "clone", repo_url], check=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE, text=True)
+        except Exception as exc:
+            print(f"Failed to clone the repository: {exc.stderr}")
+            raise
+
+        if revision:
+            subprocess.Popen(["git", "checkout", revision], cwd=str(repo_path))
+    if add_to_sys_path and str(repo_path.resolve()) not in sys.path:
+        sys.path.insert(0, str(repo_path.resolve()))
+
+    return repo_path
+
+
+def optimum_cli(model_id, output_dir, show_command=True, additional_args: Dict[str, str] = None):
+    export_command = f"optimum-cli export openvino --model {model_id} {output_dir}"
+    if additional_args is not None:
+        for arg, value in additional_args.items():
+            export_command += f" --{arg}"
+            if value:
+                export_command += f" {value}"
+
+    if show_command:
+        from IPython.display import Markdown, display
+
+        display(Markdown("**Export command:**"))
+        display(Markdown(f"`{export_command}`"))
+
+    subprocess.run(export_command.split(" "), shell=(platform.system() == "Windows"), check=True)