Published On: Tue, Jun 20th, 2017

Google releases new TensorFlow Object Detection API


Google is releasing a new TensorFlow intent showing API to make it easier for developers and researchers to brand objects within images. Google is perplexing to offer a best of morality and opening — a models being expelled currently have achieved good in benchmarking and have turn frequently used in research.

The handful of models enclosed in a showing API embody complicated avocation inception-based convolutional neural networks and streamlined models designed to work on reduction worldly machines — a MobileNets singular shot detector comes optimized to run in real-time on a smartphone.

Earlier this week Google announced a MobileNets family of lightweight mechanism prophesy models. These models can hoop tasks like intent detection, facial approval and landmark recognition.

Today’s smartphones don’t possess a computational resources of incomparable scale desktop and server-based setups, withdrawal developers with dual options. Machine training models can run in a cloud, though that adds latency and requires an internet tie — non-starters for a lot of common use cases. The choice proceed is simplifying a models themselves, creation a trade-off in a seductiveness of some-more whole deployment.

Google, Facebook and Apple have been pouring resources into these mobile models. Last fall, Facebook announced a Caffe2Go horizon for building models to run on smartphones — a initial large doing of this was Facebook’s Style Transfer. This open during I/O, Google expelled TensorFlow lite, it’s chronicle of a streamlined appurtenance training framework. And many recently during WWDC, Apple pushed out CoreML, a try to revoke a problem of using appurtenance training models on iOS devices.

Of march Google’s open cloud offerings give it differentiated positioning with honour to both Facebook and Apple, and it’s not new to delivering mechanism prophesy services during scale vis-à-vis a Cloud Vision API.

Today’s TensorFlow intent showing API can be found here. Google wants to make it additional easy to play with and exercise so a whole pack comes prepackaged with weights and a Jupyter notebook.

About the Author

Leave a comment

XHTML: You can use these html tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>