Deploy Deep Learning Application with Azure Container Instance - Devdays2018
1 like234 views
The document discusses the deployment of deep learning applications using Azure Container Instances, emphasizing the benefits of containerization such as reproducibility and agile delivery. It highlights the increasing trend of enterprises adopting containerized applications, predicts further growth by 2020, and provides practical tips on using Docker for development. Additionally, it offers resources and references for further learning on the topic.
Deploy Deep Learning Application with Azure Container Instance - Devdays2018
- 1. 利用 Container Instance 部署深度學習應用 Deploy Deep Learning Application with Azure Container Instance - Mia Chang
- 2. 利用 Container Instance 部署深度學習應用 Deploy Deep Learning Application with Azure Container Instance WHY AM I HERE... Microsoft Data Platform MVP MY JOB TITLE... Data Scientist Computer Vision/Algorithm Research AS A COMMUNITY Co-Organizer R-Ladies Taipei, Azure Taiwan, Tech Podcast Night Mia [email protected] pymia @github
- 3. WHAT BRINGS YOU HERE? Container Instance 深度學習應用 部署 Developer Data Scientist IT Pro Else Deep Learning Application Deploy
- 4. 利用 Container Instance 部署深度學習應用 那些成為資料科學家秘技之外的事。 Those Behind The Scenes... https://blog.udacity.com/wp-content/uploads/2014/11/Data-Science-Skills-Udacity-Matrix.png
- 7. Data Scientist vs Data Scientist
- 10. Tensorflow Update Pip Update Can’t Find Waldo Use Scenarios https://lh3.googleusercontent.com/Ns1ea98duQ9csaHN7EQjl6SbnLMxg_ou9bLhZA8KLqbeSv7gg7XXGX1rqtWXgnjiJjBQpiTZWyxIhtxjnWPKTM27-J9T0QI=s688
- 15. Data Scientist vs Solution Architect
- 17. Reproduce it with Docker •Docker provides you with an easy way to share your working environments including libraries and drivers. This enables us to create reproducible data science workflows.
- 18. The container advantage Fast iteration Agile delivery Immutability Cost savings Elastic bursting Efficient deployment For ITFor developers
- 19. Industry analysts agree “By 2020, more than 50% of enterprises will run mission-critical, containerized cloud-native applications in production, up from less than 5% today.”
- 22. If you need more reference …https://www.datadoghq.com/docker-adoption/
- 28. Best Time to Use Docker # Start a new project, and develop with that env # Co-work with your team # When we would like deploy our product
- 29. Demo Local Env
- 31. Demo Azure Env
- 38. Reference: Tips to build a nice Dockerfile
- 39. Microsoft Confidential FROM jjanzic/docker-python3-opencv:contrib-opencv-3.4.0 WORKDIR / RUN pip install --upgrade pip && pip install --no-cache-dir -e 'git+https://github.com/ahundt/cocoapi.git#egg=pycocotools &subdirectory=PythonAPI' && pip install --no-cache-dir requests && pip install --no-cache-dir orderedset && pip install --no-cache-dir pandas && pip install --no-cache-dir tensorflow && pip install --no-cache-dir keras && pip install --no-cache-dir dlib && pip install --no-cache-dir flask_restful && pip install --no-cache-dir psutil && pip install --no-cache-dir pymongo && pip install --no-cache-dir docker # testing RUN pip install --no-cache-dir requests-mock && pip install --no-cache-dir nose && pip install --no-cache-dir coverage && pip install --no-cache-dir nosexcover # darkflow RUN git clone https://github.com/thtrieu/darkflow.git && pip install --no-cache-dir darkflow/. && rm -rf darkflow ENV PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION python COPY . /inference RUN pip install --no-cache-dir --timeout 1000 /inference/. && rm -rf ~/.cache/pip
- 40. Microsoft Confidential Tips to build a nice Dockerfile
- 41. Microsoft Confidential Tips to build a nice Dockerfile
- 42. Tips to build a nice Dockerfile https://github.com/bcbcarl/docker-rladies-taipei