Winston Robson gumdropsteve

Tested in Ubuntu 16.04 (test machine has GPUs. may need some modification if have no GPU)

How-To-Use:

	# folder where the image is already saved (without bounding boxes)
	data_dir = 'media/stable_numbers/number_crops/'

	# folder where the new image with bounding boxes should be saved
	out_dir = 'media/copy_labeled_number_crops/'

	# go through the unique file names (because files can have more than 1 bounding box)
	for file in new_df.filename.unique():

	# load in the image and .loc[] the rows where it's referenced (each row represents 1 bounding box)

	import pynvml

	pynvml.nvmlInit()
	handle = pynvml.nvmlDeviceGetHandleByIndex(0)
	device_name = pynvml.nvmlDeviceGetName(handle)

	if (device_name != b'Tesla T4') and (device_name != b'Tesla P4') and (device_name != b'Tesla P100-PCIE-16GB'):
	raise Exception("""
	Unfortunately this instance does not have a T4, P4 or P100 GPU.\n
	Please make sure you've configured Colab to request a GPU instance type.\n

	import os
	import urllib

	def data_check(file_name, data_dir='', base_url='https://s3.amazonaws.com/nyc-tlc/trip+data/'):
	"""
	1. check if the given {file_name} exists locally in {data_dir}
	2. download from {base_url} & save in {data_dir) if not

	INPUTS
	> file_name