jgraving

#!/bin/bash

### steps ####
# verify the system has a cuda-capable gpu
# download and install the nvidia cuda toolkit and cudnn
# setup environmental variables
# verify the installation
###

### to verify your gpu is cuda enable check

jgraving

# WARNING: These steps seem to not work anymore!

#!/bin/bash

# Purge existign CUDA first
sudo apt --purge remove "cublas*" "cuda*"
sudo apt --purge remove "nvidia*"

# Install CUDA Toolkit 10
wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/cuda-repo-ubuntu1804_10.0.130-1_amd64.deb

jgraving

jgraving

# based on https://github.com/kylerbrown/ezdtw
# with modifications to be fully njit-able

import numpy as np
from numba import njit

@njit
def sqeuclidean(a, b):
    return np.sum((a - b)**2)

jgraving

#!/bin/bash
## This gist contains step by step instructions to install cuda v9.0 and cudnn 7.2 in ubuntu 18.04

### steps ####
# verify the system has a cuda-capable gpu
# download and install the nvidia cuda toolkit and cudnn
# setup environmental variables
# verify the installation
###

jgraving

Numba Ball Tree (ParallelAccelerator)

This is a modified gist of Jake Vanderplas wonderful Numba Ball Tree code from the following gist. This gist basically adds the 'nopython' parameter in the jit decorators from the original gist and parallelizes the nearest neighbor query for each of the points.

Timings

With some of the new advances in numba and the modifications

jgraving

All code is highly based on Ildoo Kim's code (https://github.com/ildoonet/tf-openpose) and derived from the OpenPose Library (https://github.com/CMU-Perceptual-Computing-Lab/openpose/blob/master/LICENSE)

jgraving

import keras.backend as K
from keras.layers import Layer
from keras.legacy import interfaces
from keras.engine import InputSpec
from keras import activations, initializers, regularizers, constraints


class DenseTransposeTied(Layer):

    @interfaces.legacy_dense_support

jgraving

import numpy as np
import subprocess as sp
import os
DEVNULL = open(os.devnull, 'w')

# load_audio can not detect the input type
def ffmpeg_load_audio(filename, sr=44100, mono=False, normalize=True, in_type=np.int16, out_type=np.float32):
    channels = 1 if mono else 2
    format_strings = {
        np.float64: 'f64le',

jgraving

#!/usr/bin/env python

import numpy as np


def medfilt (x, k):
    """Apply a length-k median filter to a 1D array x.
    Boundaries are extended by repeating endpoints.
    """
    assert k % 2 == 1, "Median filter length must be odd."