5 ms·

Deepstack: Open-Source AI Server

66 points by mimsee 5y ago

billconan 5y ago
I'm exploring torch serve and aws multi-model server. I'd like to look into this too. meanwhile, I found this one-liner confusing? > The World's Leading Cross Platform AI Engine for Edge Devices is it a server or does it run on an edge device?
- mimsee 5y ago
  It can be run on arm64 among others.
- irthomasthomas 5y ago
  Looks like it is inspired by RedisEdge https://redis.com/redis-enterprise/more/redis-edge/ https://redis.com/redis-enterprise/more/redis-edge/
  - OlafenwaMoses 5y ago
    While they may share a few similarities, DeepStack was inspired by John Olafenwa's idea to develop a unified, cross-platform AI engine that abstracts the process of using technical intensive tools and process in creating and consuming AI models, yet providing almost same level of flexibility for customizing AI for domain specific use cases, mostly in the realm of Deep Neural Network models.
- OlafenwaMoses 5y ago
  Good question: It's a server that runs on the edge and both cloud machines (e.g VMs) if you want to. Once you install an run it on your local machine (can be a laptop/desktop machine, Raspberry Pi, Jetson) or an Ubuntu VM, it turns your machine into a local server you can upload images/data to for inference. Learn more on installing and running it in the documentation linked below https://docs.deepstack.cc/ https://docs.deepstack.cc/
mdaniel 5y ago
It took quite a bit of clicking to unearth the GitHub repo for this: https://github.com/johnolafenwa/DeepStack#readme https://github.com/johnolafenwa/DeepStack#readme and there seem to be plenty of other DeepStack-ish repos in the user's profile
- fock 5y ago
  this looks fishy as hell, just look at the "parent company" https://www.deepquestai.com/about/ https://www.deepquestai.com/about/. There are these two fine gentleman, CEO and CTO. AND THEY HAVE NO NAMES. Wait, what.
  - tiborsaas 5y ago
    They are Moses.jpg and John.jpg (based on the image names) :)
    - rad_gruchalski 5y ago
      They both work at Microsoft, apparently: https://github.com/OlafenwaMoses https://github.com/OlafenwaMoses, https://github.com/johnolafenwa https://github.com/johnolafenwa.
      - OlafenwaMoses 5y ago
        Thanks for this comment to clear things up. We are currently working on updating our company's websites and fixing some of the issues in them.
kkielhofner 5y ago
Always great to see solutions for actually deploying ML models (especially incorporating edge)! I see that it's cross platform but I wonder how it stacks up to something like Triton at scale/more complex use cases? https://developer.nvidia.com/nvidia-triton-inference-server https://developer.nvidia.com/nvidia-triton-inference-server
gwern 5y ago
'DeepStack' is already the name of an important major DL poker playing algorithm.
OlafenwaMoses 5y ago
Hi everyone, I am Moses Olafenwa, the CEO and Head of Operations of the company DeepQuest AI (now a UK company based in Greenwich, London). First, apologies for the missing information of our names on the about page. The deepquestai.com website hasn't been updated recently due to our team being a small one and having lots to do. About DeepStack: DeepStack is an artificial intelligence server we developed late 2018 to empower developers to easily setup, integrate and leverage AI functionalities (blog post: https://medium.com/deepquestai/deepstack-build-ai-powered-applications-offline-c115905729b6 https://medium.com/deepquestai/deepstack-build-ai-powered-ap... ) fully on the edge or their private cloud machines. Today, DeepStack is available as a Docker image on Docker Hub and native application for Windows, with support for CPUs, modern NVIDIA GPUs, Jetson devices, ARM devices, Linux and MacOS with over 10 millions installs on Docker Hub ( https://hub.docker.com/r/deepquestai/deepstack https://hub.docker.com/r/deepquestai/deepstack ). It include functionalities such as Face Recognition/Detection APIs, common objects detection, image superresolution, custom detection models to detect any custom object of interest and many more. You can learn more about the product and using it via the documentation linked below https://docs.deepstack.cc https://docs.deepstack.cc DeepStack has a very active community on the official forum https://forum.deepstack.cc https://forum.deepstack.cc and other forums like HomeAssistant ( https://community.home-assistant.io/t/face-and-person-detection-with-deepstack-local-and-free/92041 https://community.home-assistant.io/t/face-and-person-detect... ), IPCamTalk( https://ipcamtalk.com/tags/deepstack/ https://ipcamtalk.com/tags/deepstack/ ) and YouTube ( https://www.youtube.com/results?search_query=deepstack+ai+server https://www.youtube.com/results?search_query=deepstack+ai+se... ) From year 2019 after we released DeepStack till early, 2021 the project was largely developed and maintained by John Olafenwa (the creator of the AI server) and I; pretty hard for 2 folks with day job to keep up with everything involved in the project which is why some of our sites are not fully updated. That is about to change in 2022 as we grew to a team of 6 late last year and growing to ensure we keep developing, maintaining and improving DeepStack for the almost a million developers leveraging the server to integrate AI.

66 points by mimsee 5y ago

I'm exploring torch serve and aws multi-model server. I'd like to look into this too. meanwhile, I found this one-liner confusing? > The World's Leading Cross Platform AI Engine for Edge Devices is it a server or does it run on an edge device?

mimsee 5y ago
It can be run on arm64 among others.
irthomasthomas 5y ago
Looks like it is inspired by RedisEdge https://redis.com/redis-enterprise/more/redis-edge/ https://redis.com/redis-enterprise/more/redis-edge/
- OlafenwaMoses 5y ago
  While they may share a few similarities, DeepStack was inspired by John Olafenwa's idea to develop a unified, cross-platform AI engine that abstracts the process of using technical intensive tools and process in creating and consuming AI models, yet providing almost same level of flexibility for customizing AI for domain specific use cases, mostly in the realm of Deep Neural Network models.
OlafenwaMoses 5y ago
Good question: It's a server that runs on the edge and both cloud machines (e.g VMs) if you want to. Once you install an run it on your local machine (can be a laptop/desktop machine, Raspberry Pi, Jetson) or an Ubuntu VM, it turns your machine into a local server you can upload images/data to for inference. Learn more on installing and running it in the documentation linked below https://docs.deepstack.cc/ https://docs.deepstack.cc/

mdaniel 5y ago

It took quite a bit of clicking to unearth the GitHub repo for this: https://github.com/johnolafenwa/DeepStack#readme https://github.com/johnolafenwa/DeepStack#readme and there seem to be plenty of other DeepStack-ish repos in the user's profile

fock 5y ago
this looks fishy as hell, just look at the "parent company" https://www.deepquestai.com/about/ https://www.deepquestai.com/about/. There are these two fine gentleman, CEO and CTO. AND THEY HAVE NO NAMES. Wait, what.
- tiborsaas 5y ago
  They are Moses.jpg and John.jpg (based on the image names) :)
  - rad_gruchalski 5y ago
    They both work at Microsoft, apparently: https://github.com/OlafenwaMoses https://github.com/OlafenwaMoses, https://github.com/johnolafenwa https://github.com/johnolafenwa.
    - OlafenwaMoses 5y ago
      Thanks for this comment to clear things up. We are currently working on updating our company's websites and fixing some of the issues in them.

kkielhofner 5y ago

Always great to see solutions for actually deploying ML models (especially incorporating edge)! I see that it's cross platform but I wonder how it stacks up to something like Triton at scale/more complex use cases? https://developer.nvidia.com/nvidia-triton-inference-server https://developer.nvidia.com/nvidia-triton-inference-server

gwern 5y ago

'DeepStack' is already the name of an important major DL poker playing algorithm.

OlafenwaMoses 5y ago

Hi everyone, I am Moses Olafenwa, the CEO and Head of Operations of the company DeepQuest AI (now a UK company based in Greenwich, London). First, apologies for the missing information of our names on the about page. The deepquestai.com website hasn't been updated recently due to our team being a small one and having lots to do. About DeepStack: DeepStack is an artificial intelligence server we developed late 2018 to empower developers to easily setup, integrate and leverage AI functionalities (blog post: https://medium.com/deepquestai/deepstack-build-ai-powered-applications-offline-c115905729b6 https://medium.com/deepquestai/deepstack-build-ai-powered-ap... ) fully on the edge or their private cloud machines. Today, DeepStack is available as a Docker image on Docker Hub and native application for Windows, with support for CPUs, modern NVIDIA GPUs, Jetson devices, ARM devices, Linux and MacOS with over 10 millions installs on Docker Hub ( https://hub.docker.com/r/deepquestai/deepstack https://hub.docker.com/r/deepquestai/deepstack ). It include functionalities such as Face Recognition/Detection APIs, common objects detection, image superresolution, custom detection models to detect any custom object of interest and many more. You can learn more about the product and using it via the documentation linked below https://docs.deepstack.cc https://docs.deepstack.cc DeepStack has a very active community on the official forum https://forum.deepstack.cc https://forum.deepstack.cc and other forums like HomeAssistant ( https://community.home-assistant.io/t/face-and-person-detection-with-deepstack-local-and-free/92041 https://community.home-assistant.io/t/face-and-person-detect... ), IPCamTalk( https://ipcamtalk.com/tags/deepstack/ https://ipcamtalk.com/tags/deepstack/ ) and YouTube ( https://www.youtube.com/results?search_query=deepstack+ai+server https://www.youtube.com/results?search_query=deepstack+ai+se... ) From year 2019 after we released DeepStack till early, 2021 the project was largely developed and maintained by John Olafenwa (the creator of the AI server) and I; pretty hard for 2 folks with day job to keep up with everything involved in the project which is why some of our sites are not fully updated. That is about to change in 2022 as we grew to a team of 6 late last year and growing to ensure we keep developing, maintaining and improving DeepStack for the almost a million developers leveraging the server to integrate AI.