v

annoy (approximate nearest neighbor)

1

not really an issue, but a question/comment as id love to contribute to phistoffury... have you looked into piggybacking off spotify's annoy framework in lieu of kd trees? https://github.com/spotify/annoy

layer

1

according to this: `features = model.predict(frames_arr, batch_size=batch_size)`, you are extracting the output layer as feature representations. (in which case, thats incorrect: as youd want the last fully connected layer -...

include turtlebot instructions following ros installation!