matteobarbieri

#!/bin/bash

####################
### INSTRUCTIONS ###
####################

# 1. To speed up things, switch to bt mode and select the right channel (it will make things easier in the next steps)
# 2. Switch back to USB mode (so that you can actually enter commands)
# 3. Run this script and then quickly switch to bt mode and once the blinking stops hold FN + the corresponding bt channel;
# for instance, I have paired it with Windows on channel 1, and on Linux on channel 2, so in my case it's FN + 2. Release it

matteobarbieri

import torch


def smart_classify(
        resized, net, threshold=75, device='cuda',
        conf_threshold=0.9, debug=False):
    """
    First determined whether the cell is empty or not, checking the number of
    pixels different from zero.
    Then use a neural network to classify digits.

matteobarbieri

from torch.utils.data import Dataset

import torch

from PIL import Image
from PIL import ImageFont
from PIL import ImageDraw

import glob
import random

matteobarbieri

def find_empty_grid(x):  # checks for empty spaces '0'.
    for i in range(len(x)):
        for j in range(len(x[0])):
            if x[i][j] == 0:
                return (i, j)  # return row, col

    return None


def solve_array(x):

matteobarbieri

import cv2
import numpy as np

def crop_and_warp(img: np.ndarray, crop_rect):
    """
    Crops and warps a rectangular section from an image into a square of
    similar size.
    """

    # Rectangle described by top left, top right, bottom right and bottom left

matteobarbieri

import cv2
import numpy as np
import operator


def find_corners_of_largest_polygon(img: np.ndarray):
    """Finds the 4 extreme corners of the largest contour in the image."""

    # Find contours
    contours, h = cv2.findContours(

matteobarbieri

import cv2
import numpy as np

def pre_process_image(img: np.ndarray, skip_dilate: bool=True):
    """
    Uses a blurring function, adaptive thresholding and dilation to expose
    the main features of an image.
    """

    # Gaussian blur with a kernal size (height, width) of 9.

matteobarbieri

import numpy as np

from scipy.signal import convolve2d


def print_board(board):
    R, C = board.shape
    for i in range(R):
        row = ""
        for j in range(C):

matteobarbieri

git clone https://github.com/CSSEGISandData/COVID-19
cd COVID-19
conda create --name jupyter python=3 matplotlib seaborn \
    pandas scikit-learn
conda activate jupyter
conda install -c conda-forge jupyterlab
mkdir notebooks
jupyter lab