Deebul Nair deebuls

Deep learning workshop materials

Udemy Course Liked the topics which are covered and the handson examples used
CS231n: Convolutional Neural Networks for Visual Recognition With video lectures here
Thories of Deep Learning

Navigate to http://nodejs.org/download/ and on the Linux Binaries (.tar.gz) row click to download the 64-bit version of the current latest release.
Say you've downloaded node-v0.10.7-linux-x64.tar.gz into the Downloads directory. Then, open the terminal and type the following:

$ cd ~/Downloads
$ mkdir -p ~/local/node
$ tar xzf node-v0.10.7-linux-x64.tar.gz -C ~/local/node --strip-components=1
$ echo '# Node Enviroment Setup' >> ~/.bashrc
$ echo 'export PATH=$HOME/local/node/bin:$PATH' &gt;&gt; ~/.bashrc

	---Utility for keymap creation.
	---@param lhs string
	---@param rhs string\|function
	---@param opts string\|table
	---@param mode? string\|string[]
	local function keymap(lhs, rhs, opts, mode)
	opts = type(opts) == 'string' and { desc = opts }
	or vim.tbl_extend('error', opts --[[@as table]], { buffer = bufnr })
	mode = mode or 'n'
	vim.keymap.set(mode, lhs, rhs, opts)

	import math
	import torch

	"""
	Differentiable dual quaternion distance metric in PyTorch.

	Acknowledgements:
	- Function q_mul(): https://github.com/facebookresearch/QuaterNet/blob/main/common/quaternion.py
	- Other functions related to quaternions: re-implementations based on pip package "pyquaternion"
	- Functions related to dual quaternions: re-implementations based on pip package "dual_quaternions"

	/*
	This is an example of how to use the PCL filtering functions in a real robot situation.

	This node will take an input depth cloud, and
	- run it through a voxel filter to cut the number of points down (in my case, from ~130,000 down to 20,000)
	- with a threshold to remove noise (requires minimum 5 input points per voxel)
	- then transform it into the robot footprint frame, i.e. aligned with the floor
	- subtract the robot footprint (our depth sensor's FOV includes the robot footprint, which changes dynamically with the load)
	- subtract points in the ground plane, with different tolerances for near or far
	- ground plane by default z = 0, but

	find . -name "*.zip" \| while read filename; do unzip -o -d "`dirname "$filename"`" "$filename"; done;
	find . -name ".tar" \| while read filename; do tar xvf "$filename" -C "`dirname "$filename"`"; done;

	find . -name "*.ipynb" -execdir jupyter nbconvert --to html {} \;