Roast our machine learning platform – TrainingSet.AI

[roy32]

Hello, fellow roasters!

I’m excited to finally share with you a start-up project my team and I have been working on full steam ahead, and I’m eager to see how good of a roast you can give us.

TrainingSet.AI is a platform whose primary objective is bringing a cohesive tool with which annotation teams / data scientists can annotate or classify the images, videos, text and 3D scenes to feed their AI models with valuable knowledge.

The platform could be of appliance on anything that needs human intervention or semi-intervention: robotics, self-driving cars, drones, image classification, content moderation, sentiment analysis and e-commerce tagging, to name a few.

Annotation and classification processes can be tedious and time-consuming tasks. Therefore, we are introducing a simple way of managing the above said procedures with an AI as the main assistant.

If you’re psyched about our project as much as we are, you can check TrainingSet.AI, create a free platform account and build a better AI application. And, please, give us a good roast!

We made a huge effort in the development of our platform, so any constructive feedback and sharing on social media will be more than welcome.

Esteban Salsano,

CEO at TrainingSet.AI

P.S. Currently available features are only one piece of the pie. We’ll be soon releasing text and video annotation. Stay tuned.

 

[hayley368]

@roy32 Have you found anyone with all of these characteristic:
The budget for this tool; a team of AI people able to do anything with the labeled data; a large image/video/text set in need of labelling; and a team of trained labelers standing by?

That seems like an unlikely buyer. I think you need to eliminate at least one of those buyer filter.

Either 1: Staff your own labelers. Your service is now taking in an unlabeled set, plus labeling instructions, and retuning a fully labeled data set wizard of oz style. Your business is about scalable labeling. Lots of good "shit shoveling" in that that Id love to pay someone else to do for me.

Or 2: Make an online labeling tool that makes decisions about which images to show labelers and uses your proprietary ML to continuously find only the edge cases to show the human. I don't need an ML team to use your tool. Just labelers for edge cases and the data. Your service is not embedded in my product labeling the easy cases with your ML and bubbling the hard cases to my labeling staff. I also don't need much data to get started.

 

[roy32]

@hayley368 Very interesting feedback and point of view/process and the client picture. Will work on it on the next days

 

[613jono]

@roy32 Calling yourself CEO even before a working MVP. The road ahead looks tough for you.

 

[roy32]

@613jono Thanks! The MVP is ready , good if you can check it out and shoot some feedback.

Yep company titles are an issue, shoot ideas!