t4ai-signature-detect-server - v1.0.0

🚀 Features

• Initial Deploy and Serve scripts
• Inference with request, tritonclient and ultralytics. Gradio interface for test
• Model Repository for Triton local testing
• Add Dockerfile for Triton server setup with dependencies and configuration
• Enhance Dockerfile with multistage build and custom dependencies for Triton server
• Refactor Triton model configurations and update inference pipeline for ensemble processing
• Add requirements.txt with necessary dependencies for the project
• Implement GCS file download and ONNX model export functionality
• Enhance Triton server setup with Google Cloud credentials and improved model readiness check
• Clean up exportmodel.py and improve logging in inferencepipeline.py
• Add example environment configuration file
• Add Roboflow dataset download script and update .gitignore
• Add YOLOv8 inference class for object detection with ONNX Runtime
• Dockerfile.dev and docker-compose files, add triton admin key secret
• Add onnxruntime and opencv-python dependencies; update default model and image paths in inference script
• Implement compression test pipeline with concurrent processing and logging
• Update image path retrieval to support multiple dataset splits
• Update .env.example with Roboflow API key and modify .gitignore for additional files
• Add tqdm and matplotlib to requirements.txt
• Implement compression test pipeline with concurrent processing and logging
• Load environment variables from .env file in download_roboflow.py
• Plot metrics and update gradio gui
• Adicionar barra de progresso ao pipeline de inferência e otimizar a função de execução
• Adicionar os parâmetros de conf, iou threshold em FP16 para o pós processamento no ensemble model do Triton
• Enable gRPC endpoint and port 8001
• Refactor the inference interface to support uploading image folders and adjust file paths
• Batch configuration and performance parameters in pre- and post-processing models; openvino backend for yolov8 model
• Enhance the Triton client to support different schemas HTTP/gRPC
• Add support for confidence thresholds and IOU in compressed request
• Add function to draw bounding boxes and display inference results in automatic pipeline
• Atualizar a versão do Gradio para 5.12.0 no arquivo requirements.txt
• Add pull request template for consistent contributions
• Add CITATION.cff file for proper citation and metadata
• Add CODEOFCONDUCT.md to establish community guidelines
• Add issue templates for bug reports, feature requests, and code improvements
• Add SECURITY.md to outline vulnerability reporting procedures
• Add type annotations and documentation to the SignatureDetector class methods
• Update inference GUI with new Triton server URL and enhance documentation
• Add new issue templates using yml
• Add CODEOWNERS file to define repository ownership
• Make confidence and IoU thresholds optional in model configurations
• Change data types to FP16 in model settings in TritonClientPredictor
• Add command-line argument for Triton server URL in inference GUI
• Gradio Demo video
• Add training notebook

🐛 Bug Fixes

• Add roboflow dependency to requirements.txt
• Remove dependecy
• Correct roboflow dependency to requirements.txt
• Upload gradio demo via github
• Adjust validations of mandatory fields in the bug report template
• Adjust validations of mandatory fields in the bug report template
• Update issue templates to enable blank issues and adjust validation fields
• Update pull request template for improved clarity and optional fields
• Parameterize location and model artifacts repository in deployment script
• Eliminate sensitive variables and outputs
• Fix pipeline variable exposition
• Update labels and titles to English in plots
• Translate metrics to English in the inference pipeline
• Update title in CITATION.cff and remove address details

🚜 Refactor

• Details for relative dirs
• Remove unused inference and pre/post processing code
• Streamline inference pipeline and update predictor selection mechanism
• Lower score threshold for NMS filter results
• Reorganize model export and GCS upload scripts; remove deprecated files
• Enhance model preprocessing class with improved documentation and code organization
• Reorganize inference code for improved readability and structure
• Enhance predictors with improved documentation and type hints
• Add init files in modules and reorganize directories
• Separation of classes for the interface and file organization
• Remove old GCS upload/download scripts and consolidate cloud download functionality
• Remove GCP upload/download functions and enhance model export functionality
• Consolidate image paths for compression tests and remove tests comments
• Updating docstrings, comments and types
• Improve docstrings, language and enhance inference metrics

📚 Documentation

• Model Analyzer results
• Update README.md with project overview, structure, features, and installation instructions
• Update README with Inference and Ensemble Model sections and diagrams
• Fix badge formatting in README for better display
• Update README with new directory structure and installation instructions
• Update README to clarify environment setup and add endpoint access restrictions
• Atualizar a documentação do README com melhorias na estrutura e novos links para acesso aos scripts
• Update README with links to model & dataset card and enhance utils section
• Add CONTRIBUTING.md to guide new contributors
• Add @samuellimabraz as a contributor
• Add @jwillians as a contributor
• Add @NixonMSilva as a contributor
• Add @ronaldobalzi-tech4h as a contributor
• Include contributors; how to use utility scripts, and improve the overall documentation of all modules
• Update Contributing link in TOC
• Update README with placeholder values for cloud storage commands
• Fix anchor links in README

⚙️ Miscellaneous Tasks

• Update .gitignore to exclude Vertex AI test directory
• Add deployment pipeline to azure for dev environment
• Add staging and prod pipelines for azure deployment
• Remove file automatically created by cloud provider

Full Changelog: https://github.com/tech4ai/t4ai-signature-detect-server/commits/v1.0.0

- Jupyter Notebook
Published by samuellimabraz about 1 year ago