Joao P C Bertoldo jpcbertoldo

Anomaly Segmentation Metrics for Anomalib — GSoC 2023 @ OpenVINO

Google Summer of Code 2023

Project: Anomaly Segmentation Metrics for anomalib

Mentors: @samet-akcay, @djdameln

	# %%
	# Scratch of spro on a batch of images

	import torch

	# ---- Dummy data ----

	# a batch of predictions for two images
	# shape: (2, 1, 256, 256)
	predictions = torch.rand(2, 1, 256, 256)

	"""
	Source of the idea: https://stackoverflow.com/a/58491045/9582881
	"""

	import numpy as np


	def round_significant_digits(x, dtype=np.float64, precision=4):
	string = np.format_float_positional(x, unique=False, fractional=False, trim='k', precision=precision)
	return dtype(string)

	lightning_logs/
	.cache