Image Generation Using Python:

Project Description:

This project  involves in setting up a machine learning pipeline that utilizes AWS services for storage (S3) and queueing (SQS), along with GPU-accelerated computing resources (RTX A6000 pod) for running computer vision models. The specific model being used is related to stable diffusion and regularization techniques for image processing, likely involving techniques from deep learning and generative models. The final goal seems to be the execution of the main.py file, which presumably triggers the model training or inference process.

AWS services required:

Amazon S3 (Simple Storage Service):

Use: S3 is used to store data, including model checkpoints, training images, and any other relevant files needed for the machine learning pipeline. It provides scalable, durable, and highly available object storage, making it suitable for handling large datasets and model artifacts.

Amazon SQS (Simple Queue Service):

Use: SQS is employed as a message queue to facilitate communication between different components of the system, such as triggering model training or inference tasks. The FIFO (First-In-First-Out) queue type with content-based deduplication ensures that messages are processed in the order they are received and prevents duplicate messages from being processed, ensuring reliable and orderly execution of tasks.

AWS IAM (Identity and Access Management):

Use: IAM is utilized for managing access to AWS services securely. An IAM user is created with specific access policies granting permissions to interact with S3 and SQS resources. This ensures that only authorized users or services can access the necessary resources, enhancing security and compliance.

RunPod:

Use: RunPod is utilized as the deployment environment for the machine learning model. It offers secure cloud infrastructure, possibly with specialized hardware configurations like the RTX A6000 pod mentioned in the project setup. This infrastructure is crucial for running computationally intensive tasks efficiently, especially those involving deep learning models that benefit from GPU acceleration.

Python:

 Use: Python is the primary programming language used for implementing the machine learning pipeline and executing various scripts in the project. It's used for tasks such as:

-Cloning Git repositories (git clone commands).

-Installing dependencies (pip install commands).

-Executing Python scripts (python commands).

-Configuring and updating credentials and variables files.

-Running the main machine learning pipeline (execute_pipeline.py, main.py, etc.).

Execution process:

Setup AWS

• Go to AWS.
• Go to S3 and create an S3 bucket.
• Go to SQS and create a FIFO queue.
• Go to your queue settings and select the option 'Content-based deduplication'.
• Create an IAM user and attach the policy s3_sqs_access.json from this repository.
• Create access keys for the user.

Setup RunPod

• Go to RunPod.
• Go to secure cloud and launch an RTX A6000 pod.
• Select template RunPod Stable Diffusion. Unselect Start Jupyter Notebook.
• SSH into your pod.
• Execute these commands:Explaingit clone https://github.com/JoePenna/Dreambooth-Stable-Diffusion wget https://huggingface.co/panopstor/EveryDream/resolve/main/sd_v1-5_vae.ckpt apt install zip -y mkdir Dreambooth-Stable-Diffusion/training_images mv sd_v1-5_vae.ckpt Dreambooth-Stable-Diffusion/model.ckpt git clone https://github.com/djbielejeski/Stable-Diffusion-Regularization-Images-person_ddim.git mkdir -p Dreambooth-Stable-Diffusion/regularization_images/person_ddim mv -v Stable-Diffusion-Regularization-Images-person_ddim/person_ddim/*.* Dreambooth-Stable-Diffusion/regularization_images/person_ddim/ cd Dreambooth-Stable-Diffusion pip install -e . pip install boto3 pip install pytorch-lightning==1.7.6 pip install torchmetrics==0.11.1 pip install -e git+https://github.com/CompVis/taming-transformers.git@master#egg=taming-transformers pip install captionizer
• Download the files execute_pipeline.py, credentials.py, variables.py and prompts.py from this repository.
• Go to credentials.py and update it with the access keys credentials you created.
• Go to variables.py and update it with the name of your S3 bucket and the URL of your SQS queue.
• Execute the filepython execute_pipeline.py

python app

• Clone this repository.
• Install requirements.
• Go to credentials.py and update it with the access keys credentials you created.
• Go to variables.py and update it with the name of your S3 bucket and the URL of your SQS queue.
• Execute main .py GitHub Document

Under The Guidelines of :

Dr.Kavitha Sadam

Details:

P Pranav Teja

ID:2100032273