t04glovern/dog-bark-detection

Dog Bark Detection

Conda Environment

conda env create -f environment.yml
conda activate dog-bark

Basic Exploration

If you're just looking to play around with the ML model without worrying about the AWS cloud components I'd strongly suggest looking at notebook.ipyn after running the conda commands. It's set up to train the model from scratch and run inference over .wav files.

Run Prediction

QUEUE_NAME=<queue_name> TABLE_NAME=<table_name> python3 main.py

Docker

docker build -t dog-bark .
docker run --env QUEUE_NAME=<queue_name> --env TABLE_NAME=<table_name> dog-bark

ECS

# Login
aws ecr get-login-password --region ap-southeast-2 | docker login --username AWS --password-stdin 123456789012.dkr.ecr.ap-southeast-2.amazonaws.com/dog-bark-detection

# Build
docker build -t dog-bark-detection .
docker tag dog-bark-detection:latest \
    123456789012.dkr.ecr.ap-southeast-2.amazonaws.com/dog-bark-detection:latest
# Push
docker push 123456789012.dkr.ecr.ap-southeast-2.amazonaws.com/dog-bark-detection:latest

Kubernetes

gcloud auth login
gcloud config set project <PROJECT>

docker build -t gcr.io/<PROJECT>/bark-detector:latest .
docker push gcr.io/<PROJECT>/bark-detector:latest

# When connected to Kubernetes
kubectl apply -f deploy.yml

Attribution

• Sound Classification using Deep Learning

dog-bark-detection's Issues

Is there a simple, end-user version of this?

I've been looking all over the Internet for an application that I can run that will parse an audio file and extract everything it thinks is a dog bark along (I want to monitor the same dog and have plenty of samples) with the audio timestamp. Graphs would be great showing average frequency of barking per hour/day, but minimally, just a recording of all the captured dog barks would work.

I've looked at this project, and it does appear an awful lot like what I'm looking for, but I don't see anything here that shows me HOW to use this. Is there a step by step guide to using this for someone who isn't an audio specialist?

TABLE_NAME & QUEUE_NAME ?

Hi there,
I am trying to run your bark detector (Conda).
Sorry for what is probably a basic question but what are the environment variables in the title for ? to specify a mean to send messages when a barking is detected ? How to use them ?
Thanks for your code !
Jeff