Skip to content

Acoustic event detection using yamnet model. Model is deployed using tensorflow serving in docker container and Flask API

Notifications You must be signed in to change notification settings

aabeing/Event_Detection_Tensorflow_Serving

Repository files navigation

Event_Detection_Tensorflow_Serving

Acoustic event detection using yamnet (Pre-trained classifer on 521 sound classes). A flask app is created for the model and dockerized using docker container image tensorflow serving. The output graph for top_n classes is scaled on milliseconds time. Based on the output obtained any classes can be chosen and if it belongs to a top probability class then the video/audio is clipped and output file is stored. The input audio/video can be either uploaded directly from the system (Takes very less processing time) or downloaded from a link provided as input text. Youtube_dl python is used for downloading from the net. The yamnet reads input as waveforms obtained from wav files and runs the prediction much faster than the actual duration of audio.

app_test.py: Dockerized implementation(model_test.py) of flask api

About

Acoustic event detection using yamnet model. Model is deployed using tensorflow serving in docker container and Flask API

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published