Newsletter

Get the latest news from Instill AI: open source updates, community highlights, blog posts, useful tutorials and more!

Dec 31 2022
Hey ,

It has been a tremendous year for Instill AI. VDP has grown from an idea in our minds into an open-source project used and loved by our community. Here are some numbers:
  • 🌱 5 AI tasks supported
  • πŸ”Œ 40+ Pre-built data connectors
  • ⚑️380,000+ trigger inferences
  • ⭐ 600+ stars on GitHub
  • ο»ΏπŸ‘‹ο»Ώ 160+ Community members
⭐️ Help VDP reach 1K

πŸš€ Product Updates

VDP: officially renamed to Versatile Data Pipeline

Ever since we launched VDP in August, we've been actively collecting feedback from our community. Along the way we've realised that as a general ETL infrastructure, VDP is in fact capable of processing all kinds of unstructured data.

We should not limit its usage to only visual data but to more versatile data. In addition, the term, Data Preparation, has been misleading for users often thinking it has to do with only data labelling, cleaning, or wrangling. In our vision, while VDP should involve data preparation in its MLOps practice, it should not be conceptually confined to only data preparation. VDP does more than that and is focused on the overall effectiveness of the unstructured data ETL with a data-centric paradigm. In spite of the high complexity of the underneath data processing, the end form of a VDP infrastructure is nothing but a data pipeline. The term Data Pipeline is more precise to capture the core concept of VDP, hence renamed as "Versatile Data Pipeline".

New AI task: Instance Segmentation

VDP now supports the instance segmentation task! It is a Vision task to detect and delineate multiple objects of pre-defined categories in an input image.

To build an instance segmentation data pipeline, check out the Streamlit example and import the Mask R-CNN model from the GitHub repo instill-ai/model-instance-segmentation-dvc.

ο»ΏπŸ‘‰ο»Ώ Learn more

🍎 Community Highlights

Documentation in dark mode

Everyone agrees that the first priority of all features should be supporting dark mode for the documentation! Here we go! Click the top-right toggle to turn on/off the dark mode 🌚.
πŸš€ Join our discord community
As 2023 approaches, we can't wait to enter the new year with you and continue building the foundation of unstructured data ETL.

Happy New Year ο»ΏπŸŽ‰ο»Ώ,
Xiaofei
Co-founder & COO @ Instill AI
Sep 26 2022
Hey ,

We have many cooool VDP updates for September!

πŸš€ Product Updates

Introducing VDP: open-source visual data ETL
With many MLOps tools emerging in the last few years, we wrote a blog article to explain why VDP is the future for unstructured data ETL and how VDP differentiates from other MLOps tooling.
"Our mission is to make VDP the single point of visual data integration, so users can sync visual data from anywhere into centralised warehouses or applications and focus on gaining insights across all data sources, just like how the modern data stack handles structured data."
"Unlike many MLOps tooling, VDP is built from a data-driven perspective. Although the Vision AI model is the most critical component in a visual data ETL pipeline, the ultimate goal of VDP is to streamline the end-to-end visual data flow, with the transform component being able to flexibly import Vision AI models from different sources. Please see the detailed FAQ page."
πŸ‘‰ Continue reading
⭐️ Star VDP on GitHub
New CV Task: Optical Character Recognition (OCR)

Besides image classification, object detection and keypoint detection, VDP now supports the Optical Character Recognition (OCR) task!

It is a computer vision task to detect bounding boxes containing text and to process typed or handwritten text within each bounding box into machine readable text. To build an OCR data pipeline, try to import the model from the GitHub repository instill-ai/model-ocr-dvc.

πŸ‘‰ Learn more

🍎 Community for Vision AI Enthusiasts

VDP supports data pipelines for diverse scenarios.

⚑ SYNC mode for real-time tasks: process your data with HTTP or gRPC APIs to get results immediately, suitable for tasks that have low-latency requirements.
πŸ•“ ASYNC mode for on-demand workload: set up your data pipeline to process data on demand, so it processes data only when the trigger criteria are met.

Want to learn more? Read our tutorials to use VDP for your projects.

⚑ Build a shareable object detection application with VDP and Streamlit
When YOLOv7 was out, we built a web app to compare YOLOv4 and YOLOv7 side-by-side and deployed it online to share with the community.
In this tutorial, we demonstrated how to replicate this demo app. It shows that VDP and Streamlit are a perfect match if you work with ML/Data and would like to build Computer Vision prototypes fast to share with your team, clients or the world.

πŸ‘‰ Continue reading

πŸ•“ Build a Cow Counter Dashboard with VDP, Postgres and Metabase
Drone-based smart agriculture solutions have become increasingly popular due to numerous benefits, including deeper insights into farming assets and data-driven decision-making process for better efficiency and reduced operational cost.

These solutions usually include the following processes on a high level:
  • πŸ•ΉDeploy drones and sensors to collect livestock data
  • βš—οΈ Use AI technology to process and analyse the data for locating, counting or monitoring cattle
  • πŸ“Š Provide easy-to-understand insights and takeaways to enable ranchers to make data-driven decisions or take immediate actions when issues arise
To showcase the above processes, read on or watch our video to see how to build an object detection ETL pipeline using VDP to analyse a drone video of a cattle farm, and create a Cow Counter dashboard with Metabase that provides insights about cattle in the video footage.

πŸ‘‰ Continue reading

Want to showcase what you've built with VDP? Join and share with the community!
 
πŸš€ Join our discord community
Stay tuned!

Cheers 🍻,
Xiaofei
Co-founder & COO @ Instill AI
Aug 13 2022
Hey ,

We've been working hard on the open-source visual data ETL tool VDP for the last few months. A lot of gooooodies 🍬 🍰 🧁 to share with our community!

πŸš€ Product Updates

VDP: the future for unstructured data ETL

A few months ago, we announced that we're building VDP, an open-source data tool to streamline the end-to-end visual data processing pipeline:

  • Extract unstructured visual data from pre-built data sources such as cloud/on-prem storage, or IoT devices
  • Transform it into analysable structured data by Vision AI models
  • Load the transformed data into warehouses, applications, or other destinations
We believe VDP is the future for unstructured data ETL, where developers won't need to build their own data connectors, high-maintenance model serving platform or ELT pipeline automation tool.
Pre-built ETL data connectors

VDP integrates with Airbyte to provide open-source data connectors.

By leveraging the pre-built and ready-to-use data connectors, VDP is the single point of visual data integration. You can sync visual data from anywhere into a centralised warehouse and focus on gaining insights across all your data sources, instead of maintaining pipelines.

πŸ‘‰ Learn more
Import models made easy

VDP integrates with the best ML tools to make importing models super easy.

No need to change how you manage your models. No matter the model is on your computer, in a GitHub or Hugging Face repository, or in cloud storage managed by version control tools like ArtiVC or DVC, import & deploy it with one click πŸ–±οΈ.

πŸ‘‰ Learn more
Data pipelines for diverse scenarios

VDP supports data pipelines for diverse scenarios - no need to build from scratch

⚑ SYNC mode for real-time tasks: process your data with HTTP or gRPC APIs to get results immediately, suitable for tasks that have low-latency requirements.
πŸ•“ ASYNC mode for on-demand workload: set up your data pipeline to process data on demand or schedule, so it processes data only when the trigger criteria are met.
 

πŸ‘‰ Learn more
Standardise Computer Vision tasks

VDP solves popular vision tasks out of the box:

  • Image classification
  • Object detection
  • Keypoint detection
  • OCR (coming soon!)
  • The list is growingβ€¦πŸŒ±

For each task, you can import corresponding STOA models into VDP to use in the data pipelines. The prediction results are automatically converted to a standardised format ready for future analysis.

πŸ‘‰ Learn more
Brand new Console

The Console provides a unified, clean and intuitive user experience of VDP without writing any code.

Build end-to-end data pipelines 10x faster
With VDP, you can build an end-to-end pipeline to connect your visual data, convert them to structured insights and send the results to your database in less than 5 minutes. Click here to get started.

To learn more, please check the VDP documentation.

Give VDP a ⭐ for future updates and πŸ‘‰ join our community.

Instill Cloud - fully managed VDP (coming soon!)
☁️ We offer fully managed VDP service on Instill Cloud (coming soon!) with Team and Enterprise tiers to organisations that want to get all the power of VDP for their teams, without any hassle.
  • Painless setup
  • Maintenance-free infrastructure
  • Production-ready services
  • Start for free, pay as you grow 🌱
Access to Instill Cloud is currently by invitation only. If you're interested in the hosting service of VDP, join the waitlist and we'll contact you when we're ready.
Our mission is to make Vision AI accessible to everyone. If you are from academic groups and have any showcase of Vision AI in your research, please fill out the form to tell us about yourself and your project, and we will be in touch πŸ‘.

🍎 Community for Vision AI Enthusiasts

YOLOv4 vs. YOLOv7 demo powered by VDP
We've built this live demo to play around with VDP and the STOA object detectors. You can compare the performance of YOLOv4 and YOLOv7 side-by-side.
Interested in what’s behind the curtains? VDP provides a model inference server out of the box to serve as the foundation for building real-world vision applications.

For the demo, we built two VDP pipelines with YOLOv4 and YOLOv7 respectively. By triggering them with low-code, the inference results are sent and displayed on the demo.

#HowToGrow series for Data/AI practitioners
In the last few years, we observed the converging of Data and AI. New practices, new toolings, and even new roles with shifting responsibilities are emerging.

As Vision AI practitioners, we are optimistic and excited, but we also see the whole ecosystem in a frenzied state. How do we understand each other, find the common ground to collaborate, and move forward together in this big messy fun world?

That's why we started the #HowToGrow series where experienced practitioners share valuable lessons from their Data and AI journey ο»ΏπŸ‘ο»Ώ.
πŸ’‘ How to ο»Ώgrow as an AI Researcher?
By Xiaofei Du, Co-founder & COO @ Instill AI
In 7 years of working as an AI researcher (4-year PhD + 3 years at AI/CV startup) and 2 years of do-a-bit-of-everything (AI eng, backend, frontend, PM, etc.), here are three lessons I've learnt along the way:

❢ Learn more about data pipeline to have better control of your data
❷ Explain your work on a common ground
❸ Prove the value of your work: avoid optimal on paper, broken in reality

ο»ΏπŸ‘‰ο»Ώ Continue reading
πŸ’‘ How to ο»Ώgrow as an AI Engineer?
by Nguyễn Văn TÒm, AI Engineer @ Instill AI
After 7 years of working as a software engineer (firmware/middleware, mobile and backend application), I switched to AI engineer role 4 years ago. Here are my takeaways from a +10-year software industry journey:
​
❢ Polish your MLOps tooling skill
❷ Monitor production domain drift and iterate fast
❸ Collaborate across different roles

ο»ΏπŸ‘‰ο»Ώ Continue reading

We'd love to hear more about your story. What's your most valuable lesson? What frustrates you most in your daily work? Share your story on Twitter and tag us @instill_tech 🐝!
Stay tuned and stay well!

Cheers 🍻,
Xiaofei
Co-founder & COO @ Instill AI
Apr 04 2022
Hey ,

Buckle up, we have many exciting updates for March!

🍎 Community for Vision AI Enthusiasts

2022 Data + Vision AI Survey
How is the converging of Data and Vision AI shaping your career and daily work? Let's figure it out together to better understand how stakeholders participate and collaborate with each other in the current Data + Vision AI era.

πŸ‘‰ Check out the open survey. Your input will be greatly appreciated. We will share the survey result publicly with the community.

2022 Galaxy Summit

Our CEO Ping-Lin joined a panel in this year's Galaxy Summit with Taboola's Vice President Ning Ning Wu, Gojek's Vice President Pulkit Khanna, Canner Data's CEO Howard Chi and Eslite's Vice President Frank Zhang to talk about the challenges of unstructured data processing and Vision AI in the modern data stack.

Instill AI is dedicated to making unstructured visual data analysable because only analysable data can create value. We build the open-source VDP to streamline the end-to-end visual data processing across different stakeholders.

πŸ“Ί Watch the panel stream here.

πŸš€ Product Updates

Open-source VDP - Standardise post-processing more easily
To get the model inference output in a standardised format, check our guidelines to create a model that inherits the corresponding post-processing task class instead of creating it from scratch.
Standardise post-processing for the classification task
Standardise post-processing for the detection task
If you like VDP, don’t hesitate to give us a ⭐ on GitHub and πŸ‘‰ join our community for Vision AI Enthusiasts.


Instill Cloud tutorial - How to create a detection pipeline in minutes
The tutorial shows how to use Instill AI Cloud service to create an object detection pipeline with a ready-to-use Vision AI model to process image data via HTTP requests.
Stay well!

Cheers 🍻,
Xiaofei
Co-founder & COO @ Instill AI
Mar 05 2022
Hey ,

What a busy month. Who's ready for the monthly update of February!

πŸš€ Product Updates

Open-source visual data Preparation (VDP)

We are so excited to announce the VDP project. It is an open-source tool to streamline the end-to-end visual data processing pipeline. In a nutshell, VDP can:
  • Ingest unstructured visual data from data sources such as data lakes or IoT devices;
  • Transform visual data to meaningful structured data representations by Vision AI models;
  • Load the structured data into warehouses, applications, or other destinations.
To build an open platform, we encourage all sorts of integrations and collaborations, and would like to learn your feedback and exchange ideas! If you like the project, don’t hesitate to give us a ⭐ on GitHub and πŸ‘‰ join our community for Vision AI Enthusiasts.


Instill Cloud - with human-readable pipeline name

Instead of using a randomly generated ID, create a pipeline with a human-readable name as the unique identifier.

πŸ“– Monthly Digest

Check our official blog, where we publish useful guides and resources on visual data preparation, Vision AI and more!

Unstructured data tool to break the tech and team silos
To date, there isn’t any simple tool that can make one easily tap on the value of unstructured visual data. Read our article Missing piece in the modern data stack: visual data preparation on how visual data preparation streamlines visual data processing and eliminates the silos across different roles with a standardised framework.
That's all for now. Stay well!

Cheers 🍻,
Xiaofei
Co-founder & COO @ Instill AI
Jan 29 2022
Hey ,

Happy 2022 πŸŽ‰! It's a great way to start the new year with our product updates and interesting reads. Let's dive into the monthly update of January!

🍎 Community for Vision AI Enthusiasts

This is our biggest news! We are building a community where AI engineers, AI researchers, data engineers and data scientists share ideas of Vision AI, visual data preparation and MLOps:

  • πŸ’‘ Share articles, papers, useful tutorials and learn best practices about how to apply Vision AI to solve real-world problems
  • πŸ¦„ Find like-minded friends and partnerships for your projects
  • 😊 Help out each other with your domain expertise and use cases
  • ... and more!

πŸš€ Product Updates

Instill Cloud - with a brand new console

With Instill Cloud, you don't have to worry about managing AI infrastructure, scalability or latency. We provide
  • 🧰 CLI / SDKs tools for developers to integrate with existing data stack in minutes
  • ✨ A console for non-developers to start processing visual data without coding
  • 😊 Premium support from Instill AI team
Instill Cloud is currently in private alpha, working closely with early users to build the most effective tools for visual data preparation. If this sounds interesting to you, request early access here.
 


Instill CLI - operate in the terminal

Our official command line interface (Instill CLI) is an open-source tool that provides direct access to the APIs of Instill services. You can develop shell scripts to manage your resources. See on GitHub.
 


Better performance - support for both REST and gRPC

Besides RESTful API, we now support gRPC. With smaller message payloads and rigorous specifications, the gRPC API is built to achieve low latency and high throughput communication.

πŸ“– Monthly Digest

We have launched our official blog, where we publish useful guides and resources on visual data preparation, Vision AI and more! Check it out here.

Standardise visual data preparation
Wonder what is missing from the modern data stack? Wondering why Vision AI is still so inaccessible? Read our article Why Instill AI exists on how we bring Vision AI into the modern data stack by standardising visual data preparation.
And that's all for now. Stay well! I look forward to everything we will achieve together in 2022.

Cheers 🍻,
Xiaofei
Co-founder & COO @ Instill AI