Bohumír Zámečník bzamecnik

During filming of off-road piano improvisations in New Zealand a Phantom 4 drone crashed in the flight, the last video is corrupted. Here's how we fixed it:

DJI_0065.mov - corrupt
DJI_0065.mov - OK

Let's compare the metadata:

$ ffmpeg -i DJI_0065.mov
[...]

firmware: Lenovo_P70-A_S134_150729

There's a strange bug in Lenovo P70 which affects WiFi. Let me describe it.

Ocassionaly the WiFi diconnects and is not able to connect again. When I try to connect it asks me for password and then just shows Saved, but doesn't connect. Also the UI in the WiFi settings behaves stranglely. When I turn off WiFi it stays enabled, etc.

The solution that works sometimes is to go into the Flight mode and back. Sometimes it helps for the first time, sometimes after a few tries. Also restart works, but takes more time.

However, the flight mode affect some other features. It changes the sleep mode timer to less then 1 minute, so I need to adjust it again. Also it disables automatic screen rotation.

	#!/usr/bin/env python3
	# Extracts just a dataset of labels (ingoring audio) from the full
	# MusicNet dataset in order to shrink it from 11 GB.
	# Note: I was not able to save objects (not numpy arrays) back to npz file,
	# so we save to pickle.

	import numpy as np
	from tqdm import tqdm

	# the npz file was picked in Python 2, we're opening in Python 3

	package com.github.fge.jsonschema

	import com.fasterxml.jackson.databind.{JsonNode, ObjectMapper}
	import com.github.fge.jsonschema.main.{JsonSchema, JsonSchemaFactory}
	import org.scalatest.FunSuite

	// Example of using JSON-schema validator
	// https://github.com/java-json-tools/json-schema-validator
	//
	// Code Samples - Example 1 - translated to a Scala test:

	# Decrypt password-protected PDF in Python.
	# cleaned-up version of http://stackoverflow.com/a/26537710/329263
	#
	# Requirements:
	# pip install PyPDF2
	#
	# Usage: decrypt_pdf('encrypted.pdf', 'decrypted.pdf', 'secret_password')

	from PyPDF2 import PdfFileReader, PdfFileWriter

	"""
	Joins multiple single-file invoices in the StormWare Pohoda XML format to a
	single XML file.

	https://www.stormware.cz/xml/schema/version_2/invoice.xsd
	https://www.stormware.cz/schema/version_2/data.xsd

	Usage:
	$ python join_invoices.py 'single_invoices/*.xml' merged_invoice.xml

	# On Linux we can use native GNU time, on OSX we have to install gnu-time
	# manually, since native time doesn't measure memory.

	# http://man7.org/linux/man-pages/man1/time.1.html
	# http://braumeister.org/formula/gnu-time
	$ brew install gnu-time

	# can be called via gtime, since time is a bash command

	# %M - Maximum resident set size of the process during its lifetime, in Kbytes.

	"""
	Classifies sequences of length 10 with 20 features into 2 classes
	with a single LSTM layer with 32 neurons.

	See also a more involved example:
	https://gist.github.com/bzamecnik/dccc1c4fdcf1c7a31757168b19c827a7
	"""


	from keras.layers import Input, LSTM, Dense

	"""
	When classifying upon a sequence usually we stack some LSTM returning sequences,
	then one LSTM returning a point, then Dense with softmax activation.

	Is it possible instead to give the last non-sequential LSTM a softmax activation?

	The answer is yes.

	In this example we have 3 sequential layers and one layer producing the final result.

	____________________________________________________________________________________________________
	Layer (type) Output Shape Param # Connected to
	====================================================================================================
	input_1 (InputLayer) (None, 32, 10) 0
	____________________________________________________________________________________________________
	lstm_1 (LSTM) (None, 32, 10) 840 input_1[0][0]
	____________________________________________________________________________________________________
	add_1 (Add) (None, 32, 10) 0 input_1[0][0]
	lstm_1[0][0]
	____________________________________________________________________________________________________