Protecting indoor plants from your cats using image recognition, Raspberry Pi and GCP

viktorrobin commented 5 years ago

Abstract (2-3 lines)

Cats are known to enjoy playing with indoor plants in a destructive way, which causes a mess in addition to being a cost. We present here a system that leverages deep learning, Raspberry Pi and Google Cloud to detect cats playing with indoor plants and stop them from destroying them.

Keywords: tensorflow, cross-validation, deep learning, raspberry pi, image recognition, cloud, API, productionisation, docker container

Brief Description and Contents to be covered

The objective of the project is to detect if a cat is playing with indoor plants. A device built with Raspberry Pi is used to detect movement, take a picture, sends it to the GCP API (tensorflow model), which returns the probability of a cat playing with plants and, if yes, plays the sound of a hoover or dogs.

3 sections:

Building an image detection model using tensorflow to detect cats playing with plants: dataset collection and model training.
Productionisation of the model as an API, hosted on Google Cloud: Docker, GCP
Raspberry Pi device to detect movement, take picture and send it to GCP API and play sound

This a fun example of productionising a model on cloud that involves different areas that are related but usually not presented together.

Pre-requisites for the talk

None

Time required for the talk

30min

Link to slides

Not done yet

Will you be doing hands-on demo as well?

No

Link to ipython notebook (if any)

None

About yourself

Dr. Victor Robin - Data Science Director at dunnhumby linkedin

Are you comfortable if the talk is recorded and uploaded to PyData Delhi's YouTube channel ?

Yes

Any query ?

None

shagunsodhani commented 5 years ago

Hello, @viktorrobin Thank you for proposing a talk! The topic seems to be very interesting and engaging. I am looking forward to your slides. Are you also planning to make the code open-sourced so that people may play with it?

viktorrobin commented 5 years ago

Hi @shagunsodhani Thanks! Haven't done the slides yet. The code is mainly already on github but need some tidy up :)

viktorrobin commented 5 years ago

@shagunsodhani @MSanKeys963 First version of the slides here.

I will likely make some minor changes, but let me know if you have any feedback.

Thanks

MSanKeys963 commented 5 years ago

Please have a look @shagunsodhani.

shagunsodhani commented 5 years ago

The slides look good @viktorrobin though it seems incomplete (from slide 16 onwards) or maybe it is too slow to load on my side. Best of luck for the talk.

pydatadelhi / talks