vanvalenlab / kiosk-redis-consumer / 5

96%

push

travis-ci

Use a gRPC tensorflow-serving client (#13)

* Gut tornado and replace with a gRPC client.

* consume() calls iter_redis_keys.  each key is then passed to _consume()

* remove all processing consumers, instead use preprocess and post-process function wrappers around `consumer._process`

* improve logging and add timestamps for each statement

* move image manipulation functions into utils.py

* processing.py and utils.py fully tested

* add input tensor name and dtype as env vars and update defaul tf-port to 8500

* pre and post process comma-separated function list

* try to squeeze out batch dimension, may not work on 3D data?

* add travis.yml and add badges to README

463 of 463 new or added lines in 9 files covered. (100.0%)

313 of 569 relevant lines covered (55.01%)

0.55 hits per line