Platform Blog

AllNewsFeaturesTutorials

Mobile App for platform.ai users

News
Atin Matur
September 15, 2020

The idea behind the platform.ai mobile app is to help users test their deep learning models in real-time using their smartphones. After creating a new project on platform.ai, users can use their smartphone camera or gallery to select a picture and classify it using their own deep learning models on platform.ai.

Read More

Latest posts

Using recognition models on face attributes

Face attribute models are used for retrieval tasks like "find all actors with green eyes and curly hair" or for applications used to measure emotional states like smiling / not-smiling. We investigate model architectures, loss functions and training sets to determine the optimal combination for distinguishing facial attributes in a weakly-supervised setting. We then compared platform.ai’s default model against our new model, using a variety of projections based on linear and non-linear dimensionality reduction methods and discovered that the new model can more easily identify attributes like gender, race, age, baldness, hair color and attractiveness.

Sergei Chudov, Mostafa Gazar and Timothy Quill
October 6, 2020

New Food-101 SoTA with fastai and platform.ai’s fast augmentation search

Recently, better image classification models have tended to follow a trajectory towards deeper or wider networks or extensive test time augmentations. We will share some of the techniques of fastai v1 which allowed us to advance the State of the Art (SoTA) results for the Food-101 dataset, using transfer learning with a simple ResNet-50 architecture with minimal augmentations.

Amit Redkar, Krisztian Kovacs, Jim Bremner, Arshak Navruzyan
October 6, 2020

Finding useful augmentations with minimal use of compute

Data augmentation of images is a widely adopted technique to improve models generalization [3] by applying transformations (or sets of combined transformations) during training and optionally at test time. Often practitioners look for the best set of transformations manually, either by relying on domain expertise or making assumptions about generally useful transformations. This process of manual search for optimal augmentations can be time consuming and compute intensive.

Cristian Duguet, Adrian Hutter, Nandika Kalra, Arshak Navruzyan
October 6, 2020

Optimizing hyperparams for image datasets in fastai

Over the past few months at Fellowship.AI, our team has worked to identify optimal hyperparameters to achieve near State of the Art (SotA) performance on a variety of datasets while utilizing the fastai v1.0 framework. This research was pursued in order to optimize the hyperparameter settings of platform.ai, which seeks to bring deep learning to an extended (including non-technical) audiences.

Adrian Clarke, Brian McMahon, Pranav Menon, Kavan Patel
October 6, 2020
News

Mobile App for platform.ai users

The idea behind the platform.ai mobile app is to help users test their deep learning models in real-time using their smartphones. After creating a new project on platform.ai, users can use their smartphone camera or gallery to select a picture and classify it using their own deep learning models on platform.ai.

Atin Matur
September 15, 2020
Features

Video on platform

Platform.ai is an incredible tool that makes it easy for anyone to quickly label large datasets of images. But wouldn't it be great if platform.ai could also work with a similarly popular type of media – video? Support for video would be especially useful since it is more complicated to work with video compared to just images.

Noran Raskin and David Cato
September 15, 2020

Sign up to recieve a monthly email on the latest news!

Thank you! Your submission has been received!
Invalid Email. Please try again.
HomeProductPricingEnterpriseCompetitionsAboutBlog
Contact us
© Copyright Platform.ai inc. All rights reserved
Terms and policies