vbalnt

x  = np.array([[ 0.06169621],
       [-0.05147406],
       [ 0.04445121],
       [-0.01159501],
       [-0.03638469],
       [-0.04069594],
       [-0.04716281],
       [-0.00189471],
       [ 0.06169621],
       [ 0.03906215],

vbalnt

name	description
ncc	Normalised cross correlation
sift	SIFT [Lowe IJCV 2004]
rootsift	rootSIFT [Arandjelović & Zisserman CVPR 2012]
orb	ORB [Rublee et al ICCV 2011]
brief	BRIEF [Calonder et al. PAMI 2012]
binboost 	BinBoost [Trzcinski et al. PAMI 2013]
deepdesc 	DeepDesc [Simo-Serra et al. ICCV 2015]
liop 	LIOP [Wang et al ICCV 2011]
tfeat-margin-star	TFeat with margin loss [Balntas et al. BMVC 2016]

vbalnt

• git pull (to get other people's changes - run this before any change you make)
• make changes in your local copy of the repo
• git add * to add your changes
• git commit -m "your message with your changes here"to commit your changes
• git push to push your changes for everyone else

vbalnt

require 'cutorch'
require 'xlua'
require 'trepl'
require 'cunn'
require 'cudnn'
require 'image'
require 'nn'
require 'torch'
require 'lfs'

vbalnt

This document describes how to install the combination of 14.04 + CUDA 7.5 + Tensorflow. This combination is the easiest to install without anything like compilation from sources etc.

Download and install Ubuntu 14.04 or 16.04

http://releases.ubuntu.com/14.04/ http://releases.ubuntu.com/16.04/

Install CUDA 8

vbalnt

tests = int(sys.argv[1])
S = 32 # patch size
S2 = S / 2 # Clamp sample coordinates to this
sigma = S / 5.0 # Sampling geometry II
random.seed(42) # Make repeatable for simplicity

def random_coordinate():
    coord = int(round(random.gauss(0.0, sigma)))
    return min(max(coord, -S2 + 1), S2)+S2-1

vbalnt

# Run on GPU: THEANO_FLAGS=mode=FAST_RUN,device=gpu,floatX=float32 python mnist_siamese_graph.py
from __future__ import print_function

import sys
import os
import time
import numpy as np
import theano
import theano.tensor as T
import lasagne

vbalnt

'''Train a Siamese MLP on pairs of digits from the MNIST dataset.

It follows Hadsell-et-al.'06 [1] by computing the Euclidean distance on the
output of the shared network and by optimizing the contrastive loss (see paper
for mode details).

[1] "Dimensionality Reduction by Learning an Invariant Mapping"
    http://yann.lecun.com/exdb/publis/pdf/hadsell-chopra-lecun-06.pdf

Run on GPU: THEANO_FLAGS=mode=FAST_RUN,device=gpu,floatX=float32 python mnist_siamese_graph.py

vbalnt

#ifndef _UTILS_H_
#define _UTILS_H_
#include <stdio.h>
#include <string.h>
#include <dirent.h>
#include "vlfeat/generic.h"
#include "vlfeat/sift.h"
#include "vlfeat/imopv.h"
#include "pgm.h"
/* #include <plplot/plplot.h> */

vbalnt

/* Creation of the test array & calling of the learning func */
bintest orb_tests[n_tests];
learn_orb_tests_g2(training_data, orb_tests ,1024,32,10000);
/* fwrite_bintests(orb_tests,1024,"orb1024.descr"); */


/* Actual learning func */
void learn_orb_tests_g2(dataset data,bintest *ltests,int dims,int patch_size,int nlearn)
{
  bintest *all_tests;