p/picsell-ia-2
Open, collaborative, end-to-end Computer Vision Platform
Pierre-Nicolas TIFFREAU
Picsell.ia β€” Open, collaborative, end-to-end Computer Vision Platform
A platform allowing you to find and share your image datasets or trained models, build your computer vision project from scratch with a powerful annotation tool and track your experiments to deliver the best AI models in no time !
Replies
Pierre-Nicolas TIFFREAU
Hey Product Hunters πŸ‘‹ Build your first Computer Vision project on Picsell.ia here πŸ‘‰ https://picsellia.com At Picsell.ia, we are dedicated to provide the best tools and workflows for data scientists and to integrate it in our platform ! Our goal is to reduce data scientist's frustration and project's failure rate πŸ’ͺ It's been a year since we started developing those tools and we are already at our third release, still focused on computer vision right now. As we are a small team, it's not always easy to deliver such a complete product and to continue innovating on all the parts of an end-to-end computer vision platform. But today we are very proud to show you our fully redesigned platform, with many more features that I'm sure will accelerate your computer vision projects πŸš€ βœ… A dataset HUB to find and share your data with our community or your team βœ… A model HUB that has the same purpose but for pre-trained models βœ… Powerful image annotation tools where, for example, you can plug your AI model for pre-annotation or use our automatic segmentation tools to annotate 10x faster βœ… Dahsboards and tables to track and compare your experiments βœ… A serving engine allows you to run inference with any of your models right in the platform with a nice UI βœ… A python SDK and a CLI to connect with all your favorite tools and processes βœ… Version all your weights, datasets and checkpoints automatically thanks to our API βœ… And of course you can invite your team, manage their permissions and work together on your projects ! Soon to come : 🟩 One click training : Launch training on your data easily without having to worry about infrastructure 🟩 Deployment : Choose how many GPU you wants and we will provision a GPU-enabled cluster with an API endpoint so you can access your model in a scalable, production environment 🟩 Analytics engine : Analyze your training logs, data, results to provide you some insights on how to improve your models by tuning some parameters, annotate differently etc... I hope that I've awaken your curiosity and please remind that you can come talk to me or my team here or by sending us a nice email at contact@picsellia.com 😊