edgarriba

use std::{
    sync::{Arc, Mutex},
    time::Duration,
};

use zenoh::Wait;

#[derive(thiserror::Error, Debug)]
pub enum NodeError {
    #[error("Node error: {0}")]

edgarriba

# https://github.com/optuna/optuna-examples/blob/main/pytorch/pytorch_lightning_simple.py

"""
Optuna example that optimizes multi-layer perceptrons using PyTorch Lightning.
In this example, we optimize the validation accuracy of hand-written digit recognition using
PyTorch Lightning, and FashionMNIST. We optimize the neural network architecture. As it is too time
consuming to use the whole FashionMNIST dataset, we here use a small subset of it.
You can run this example as follows, pruning can be turned on and off with the `--pruning`
argument.
    $ python pytorch_lightning_simple.py [--pruning]

edgarriba

/* 
 * Copyright (c) 2019 Edgar Riba.
 * 
 * This program is free software: you can redistribute it and/or modify  
 * it under the terms of the GNU General Public License as published by  
 * the Free Software Foundation, version 3.
 *
 * This program is distributed in the hope that it will be useful, but 
 * WITHOUT ANY WARRANTY; without even the implied warranty of 
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 

edgarriba

import torch
from torch.autograd import Variable
from torch.autograd import Function

import numpy as np

anchor_str = 'p0\x0f?>yK\xbf\xcayY\xbf\x80l3?\x12\x13j?\xf3\xe3x?\x06\xbf+\xbf\x94\x10\t?D\xed\x8b=\xb9\xb9r>\'\xd9%?\xd1\x1fe?\'\x7f~?\xdc\xef\xa9\xbep\xe8\xbb>\xf4\xcc\x1b?\xd4\xda\xd5\xbe}\x12i\xbf\\L\x0e?\xe5:\xfe>{\xe1\xd3\xbe\x13\xbfD?-\xdb2\xbd2\xb2(\xbf\xfa<}\xbe-\x9fN?\xe1\xcfE?\xb6\x1cq\xbf\xb2C3?<\x0f%\xbf\xed\x80K\xbcE3i\xbe\xa2\xa55?E&\x96>\xe9\xb3f>N8}\xbf\xdf}c\xbfZa1>\xa8\x9c[\xbf\xc4x\x7f\xbf\x9e)\x83=\xcd%o>\\BC\xbf\xd7\xa0|?Mn\x8a\xbe%uw?\x80\xeeR?\x16\x19\xf2>\x85\x8f<>\x9aN\xf5>\x8d\xc3j\xbf\xc8E\x94\xbe\xa0/{?\xefq\xfc>\xaa\xf5S\xbfb\x1b}?J\x11U\xbf`\xadm?\x85\xa6\xc1=|\x0bv;\xfc\xbbp\xbf:p?\xbfV\x93\xc6=T\xfa\xb1>T\xe0z?j\xe2,?`\xe7x\xbf(\x95\x06\xbf\x00\x08\x7f?\x83\xcaJ\xbf\xd6$\x86\xbe,\x910?/\x08\x13\xbe\xa2J/\xbf\xfb\x90\xb3\xbdZ\n\xf2>UF\x88>))\xfc>J\x95;?\x91\x1b<?\x15\xf4\xa3\xbeV\x91x?\x01\xa8\\\xbf\xf2=\xa6\xbeG\xa2w\xbf\xf8\x13|?\xd8\xee5?\x8e\xbe|?\xc5\xae)>J\xe7c?>\xd2^

edgarriba

import torch
import numpy as np

import pickle

f = open('/home/eriba/software/pytorch/examples-edgarriba/triplet/nan_test.pkl', 'rb')

data = pickle.load(f)

a = torch.from_numpy(data['a']).cuda()

edgarriba

@startuml

class Tensor {
 + Tensor(cnn_size_t batch, cnn_size_t width, cnn_size_t height, cnn_size_t depth)
 + shape(): int[]
 + operator [](cnn_size_t index) : float_t&
 + operator [](cnn_size_t index) const : const float_t&
 + at(cnn_size_t batch, cnn_size_t width, cnn_size_t height, cnn_size_t depth) : float_t&
 + at(cnn_size_t batch, cnn_size_t width, cnn_size_t height, cnn_size_t depth) const : const float_t& 
 + ptr(cnn_size_t batch, cnn_size_t width, cnn_size_t height, cnn_size_t depth) : float_t&* 

edgarriba

@startuml
enum DeviceType {
 CPU
 GPU
}

class Layer {
}

class Context {

edgarriba

@startuml
package layers #DDDDDD {

enum Engine {
 CUSTOM
 NNPACK
 AVX
 AVX2
 SSE
 OPENCL

edgarriba

// Returns the camera matrix given the focal and the image itself
// img The reference to the image
// f The focal length in pixels of the camera
cv::Mat getCameraMatrix(const cv::Mat& img, const float f) {
  const int cx = img.cols / 2;
  const int cy = img.rows / 2;
  return (cv::Mat_<float>(3, 3) << f, 0, cx, 0, f, cy, 0, 0, 1);
}

// Convert a set of points to bearing

edgarriba

/* Reimplementation of `AddImagesToReconstructionBuilder`
   https://github.com/sweeneychris/TheiaSfM/blob/master/applications/build_reconstruction.cc#L334
   Using an external local features extractor. In this case AKAZE from OpenCV.
*/
void AddImagesToReconstructionBuilder(
    ReconstructionBuilder* reconstruction_builder) {
  std::vector<std::string> image_files;
  CHECK(theia::GetFilepathsFromWildcard(FLAGS_images, &image_files));

  // Create the feature matcher.