AutoML Vision - Part 1

Подписаться 1,2 млн

Просмотров 117 тыс.

50% 1

In this episode of AI Adventures, Yufeng Guo uses AutoML Vision to build and employ a machine learning model that recognizes different types of….chairs! In part 1, he’ll walk you through gathering the data and creating a csv file that describes the location and label for all the images in the dataset. Don’t miss part 2 to see how the model performs!
AutoML data preparation notebook → bit.ly/2KQO9Vm
Associated blog post → bit.ly/2L9rpfh
Watch more AI Adventures → bit.ly/AIAdventures
Subscribe to the Google Cloud Platform channel → bit.ly/GCloudPlatform
#AIAdventures

Наука

Опубликовано:

10 июл 2018

Ссылка:

Скачать:

Готовим ссылку...

Добавить в:

Мой плейлист

Посмотреть позже

Комментарии : 41

@bencox311 2 года назад

You answered in an 8 minute video what took me about 5 hours of scouring the internet to find and understand. I know this video is older at this point, but data labeling hasn't fundamentally changed in any way so it's still very relevant!

@jaredcheung6875 5 лет назад

Thanks for you sharing of the Google AutoML, it's great to understand even without some fundamental knowledge.

@madhubagroy 3 года назад

Fantastic series! Thank you so much!

@robindong3802 5 лет назад

Oh, this is what I am looking for. thanks for your great demo and code.

@slash7954 3 года назад

Thank you for teaching us AutoML. Your video was very good!

@sakshibharti8238 6 лет назад

Hello, I really enjoy watching videos at AI adventures. I am looking forward to watch Natural language processing videos at AI adventures. A video which can cover from scratch such as preprocessing to data till the end of prediction. Thanks Yufeng Guo for explaining everything in short and crisp way. :-)

@ivanrodriguezc 6 лет назад

Thanks for the video 😃

@UTECHPIA 5 лет назад

This video is excellent! I love how clear everything is explained, the pace was perfect. Using ffmpeg was a nice trick that I'll surely use in the future! 👍

@Supermanindia98 4 года назад

Is firebase ml vision automl is supported for Android 7 codes in Android studiob

@ianrichardson3888 3 года назад

Excellent. Thank you.

@howaboutyu 6 лет назад

this is really cool, thanks :)

@techtona7543 5 лет назад

thank you, what a great video..

@k14pc 6 лет назад

this is really cool that you guys are doing this. i've been looking into data science/ML and it's helpful to see that at least at the basic level you don't have to be a programming god to do it, and it's just really helpful to see what exactly cleaning data means and what exactly the job is. thanks! and i think you guys probably want more people filling this skillset huh hehe edit: ah i see you guys have a service to do the training on the cloud, that makes more sense

@nairod87 5 лет назад

Thanks a lot, that's perfect for a french student doing his internship ;)

@evgeniymamchenko9582 3 года назад

You can also try transfer learning (but no NAS) for image classification even without writing any code in an Android app called Pocket AutoML. It trains a model right on your phone without sending your photos to some "cloud" so it can even work offline.

@aftabnaveed 5 лет назад

In this video it says that minimum limit for each label is 10 images, but in my console it says 100 which is too much. Is that something which can easily be fixed?

@dodo-gp6ix 6 лет назад

Is it released? i couldn't find it in cloud console

@FinHacksQuants 6 лет назад

Hi YuFeng, is GoolgeCloud Auto ML capable to do what i have been live predicting the DOW JONES, SP500, NASDAQ FUTURES PRICES ? Kindly please let me know. Thanks

@ChandravijayAgrawal 2 года назад

just goto part 2

@amelbenaida9659 5 лет назад

I tried this on Colab , but quickly I get the error "Runtime crashed after using all ram". Is there something I could do ?

@krishnaemani7419 6 лет назад

Hi Yufeng.Im getting an error while running the code in description the error is "unicodespace codec can't decode bytes in position truncated \UXXXXXXXX escape.How to solve it plz help

@deeplearningpartnership 6 лет назад

Cool.

@eric-seastrand Год назад

Super informative! What was the rationale for using pandas and Jupiter? It feels overkill for this task. You could have built a csv far faster and simpler using standard posix commands (ls -al | awk some columns > your file.csv) but maybe there’s something I missed?

@adonaiarruda3634 4 года назад

how can I download this dataset?

@sonatjamesnew3679 5 лет назад

Please make video related how to integrate with android application. please provide some example video.

@lolatJESS 5 лет назад

I dont really know what I'm watching, but it makes a lot of sense. Should I quit my day job and start coding?

@kevin_kk 5 лет назад

hahah that's what I did as of 3 months ago my friend. I put a serious pause on my 6 years running music career to study Data Science & for me to now hear most work i'm to do will be automated is just a Woo-Hoo feeling. Join the team Jess, it's very much worth it!