Uploaded March 2020 | Updated November 2026, 4 days ago
I read an article the other which described how Airbnb uses computer vision and machine learning to automatically detect amenities (household objects) in their listings. I loved it so much I decided to replicate it in my own version as a machine learning project.
This week I downloaded all the subset of Open Images Airbnb focused on (all of their target classes such as bathtub, toilet and swimming pool).
I used my preprocessing functions to turn all of the images into label files which I saved to JSON's for easy loading later on.
And since I had all of the images downloaded, I started running a series of modelling experiments on 3 different classes (coffeemaker, bathtub, treehouse) tracking all of the results in Weights & Biases.
Next week, modelling gets stepped up.
All videos in this series:
Full playlist - dbourke.link/airbnbplaylist
Part 1 (overview) - youtu.be/C_lIenSJb3c
Part 2 (data exploration) - youtu.be/hTHXVfryZDo
Part 3 (data manipulation) - youtu.be/cijajMJ3m3w
Part 4 (building a small dog model) - youtu.be/ooB-ZU9iLmI
Part 5 (experiment tracking) - youtu.be/83s-OEFMgVY
Part 6 (modelling experimentation) - youtu.be/eiA3EEeGAZc
Part 7 (building a big dog model) - youtu.be/5X6A1i5wbFU
Part 8 (getting my modelling deployed) - youtu.be/smlQbh6jQvg
Part 9 (project wrap up and retrospective) - youtu.be/aBkUE5OJ8u8
Note: I am not affiliated with Airbnb nor are they involved with this series of videos in any way.
Links:
My machine learning course - dbourke.link/mlcourse
All of my notes/ideas in Notion - dbourke.link/airbnb42days
Airbnb Amenity Detection article - link.medium.com/05qPErz3a4
Weights & Biases (tool for tracking deep learning experiments) - wandb.com
Weights & Biases tutorials - wandb.com/tutorials
Get email updates on my work - dbourke.link/newsletter
Support on Patreon - bit.ly/mrdbourkepatreon
Connect elsewhere:
Web - dbourke.link/web
Quora - dbourke.link/quora
Medium - dbourke.link/medium
Twitter - dbourke.link/twitter
LinkedIn - dbourke.link/linkedin
#machinelearning #datascience #deeplearning
I read an article the other which described how Airbnb uses computer vision and machine learning to automatically detect amenities (household objects) in their listings. I loved it so much I decided to replicate it in my own version as a machine learning project.
This week I downloaded all the subset of Open Images Airbnb focused on (all of their target classes such as bathtub, toilet and swimming pool).
I used my preprocessing functions to turn all of the images into label files which I saved to JSON's for easy loading later on.
And since I had all of the images downloaded, I started running a series of modelling experiments on 3 different classes (coffeemaker, bathtub, treehouse) tracking all of the results in Weights & Biases.
Next week, modelling gets stepped up.
All videos in this series:
Full playlist - dbourke.link/airbnbplaylist
Part 1 (overview) - youtu.be/C_lIenSJb3c
Part 2 (data exploration) - youtu.be/hTHXVfryZDo
Part 3 (data manipulation) - youtu.be/cijajMJ3m3w
Part 4 (building a small dog model) - youtu.be/ooB-ZU9iLmI
Part 5 (experiment tracking) - youtu.be/83s-OEFMgVY
Part 6 (modelling experimentation) - youtu.be/eiA3EEeGAZc
Part 7 (building a big dog model) - youtu.be/5X6A1i5wbFU
Part 8 (getting my modelling deployed) - youtu.be/smlQbh6jQvg
Part 9 (project wrap up and retrospective) - youtu.be/aBkUE5OJ8u8
Note: I am not affiliated with Airbnb nor are they involved with this series of videos in any way.
Links:
My machine learning course - dbourke.link/mlcourse
All of my notes/ideas in Notion - dbourke.link/airbnb42days
Airbnb Amenity Detection article - link.medium.com/05qPErz3a4
Weights & Biases (tool for tracking deep learning experiments) - wandb.com
Weights & Biases tutorials - wandb.com/tutorials
Get email updates on my work - dbourke.link/newsletter
Support on Patreon - bit.ly/mrdbourkepatreon
Connect elsewhere:
Web - dbourke.link/web
Quora - dbourke.link/quora
Medium - dbourke.link/medium
Twitter - dbourke.link/twitter
LinkedIn - dbourke.link/linkedin
#machinelearning #datascience #deeplearning










![These TensorFlow updates are huuuuuuuuge! [Google I/O 2021] | Machine Learning Monthly May 2021
Machine learning monthly covers the latest and greatest (but not always the latest) of the machine learning field from the previous month. You can expect exciting machine learning updates as well as dancing.
ML Monthly May 2021 - https://zerotomastery.io/blog/machine-learning-monthly-may-2021/
Learn ML (beginner-friendly course) - https://dbourke.link/mlcourse
Learn Deep Learning + TensorFlow (code-first) - https://dbourke.link/tfcourse
Connect elsewhere:
Web - https://dbourke.link/web
Twitch - https://www.twitch.tv/mrdbourke
ArXiv channel (past streams) - https://www.youtube.com/channel/UCKt-z-AVy7duq6 N-michg
Get email updates on my work - https://dbourke.link/newsletter
Timestamps:
0:00 - Intro
0:24 - Google IO 2021
1:03 - Googles On-device Machine Learning Page
1:43 - TensorFlow Lite Models are now compatible with TensorFlow.js
2:28 - Create your own TensorFlow Lite models with the new Model Maker library
3:13 - TensorFlow Hub Upgrades
4:00 - TensorFlow Cloud (run your small experiments at cloud scale)
4:48 - Vertex AI (Google Clouds new one-stop-shop for everything machine learning)
5:31 - The Practitioners Guide to MLOps White Paper
6:13 - The new TensorFlow Town Square (TensorFlow Forum)
6:37 - People and AI Guidebook 2.0
7:53 - Know Your Data tool (beta)
8:32 - Where to learn more about Google IO ML & TensorFlow updates
8:57 - Next generation pose detection with MoveNet and TensorFlow.js
9:57 - Datasets and code on arXiv (thanks to paperswithcode)
10:31 - wav2vec-U (high performance speech recognition with no supervision)
12:14 - What is active learning? blog post by Roboflow
13:17 - Reproducible Deep Learning course by Simone Scardapane
14:25 - The Rise of HuggingFace by Mark Saroufim
15:06 - Outro & 🕺
#machinelearning #mlmontly These TensorFlow updates are huuuuuuuuge! [Google I/O 2021] | Machine Learning Monthly May 2021](https://i.ytimg.com/vi/jIiBaYvsM60/mqdefault.jpg)