MathematicalEngineeringDeepLearning/tikz/figure1-3-c-conv-nn-simple.tex at master · yoninazarathy/MathematicalEngineeringDeepLearning · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
\documentclass[tikz, border=50pt]{standalone}

\usepackage{graphicx}
\usepackage{color}
\usepackage{tikz}
\usepackage{geometry}
\usepackage{medl_colors}
\usetikzlibrary{calc}

% Use sans everywhere, first line for sans text and second two for sans math
\renewcommand{\familydefault}{\sfdefault}
\usepackage{sansmath}
\sansmath

\usetikzlibrary{arrows.meta,calc,positioning,shapes.arrows,3d}
\graphicspath{ {./images/} }

\begin{document}

	%\noindent\resizebox{\textwidth}{!}{
	\begin{tikzpicture}[
	Arrow/.style = {single arrow,draw,
    minimum height=8mm,
    minimum width=2mm,
    rotate=0}]
    % can also specify z axis slope eg [scale=.8, z={(-.707,-.3)}]
	%\draw[use as bounding box, transparent] (-1.8,-1.8) rectangle (17.2, 3.2);

		\newcommand{\networkLayer}[5]{
			\def\q{0.02} %parameter for filling

			\def\s{#1} % Size of square
			\def\x{#2} % X distance
			\def\y{#3} % Y distance
			\def\z{#4} % Z distance

			% NOTE: X axis passes through centre of default square (if Y and Z are both zero)
			% Construct clockwise from bottom back
			\coordinate (A) at (\x,-\s/2+\q/2+\y,-\s/2+\q+\z);
			\coordinate (B) at (\x,-\s/2+\q/2+\y,\s/2-\q+\z);
			\coordinate (C) at (\x,\s/2-\q/2+\y,\s/2-\q+\z);
			\coordinate (D) at (\x,\s/2-\q/2+\y,-\s/2+\q+\z);

			\filldraw[#5] (A) -- (B) -- (C) -- (D);  % Fill before draw edges

			\draw[line width=0.3mm, fill-blue] (A) -- (B);  % Bottom line
			\draw[line width=0.3mm, fill-blue] (B) -- (C);  % Vertical front
			\draw[line width=0.3mm, fill-blue] (C) -- (D);  % Top line
			\draw[line width=0.3mm, fill-blue] (D) -- (A);  % Vertical back

		}

		\newcommand{\pyramid}[6]{
			\def\q{0.02} %parameter for filling

			\def\s{#1} % Size of square
			\def\x{#2} % X distance
			\def\y{#3} % Y distance
			\def\z{#4} % Z distance

			% NOTE: X axis passes through centre of default square (if Y and Z are both zero)
			% Construct clockwise from bottom back
			\coordinate (A) at (\x,-\s/2+\q/2+\y,-\s/2+\q+\z);
			\coordinate (B) at (\x,-\s/2+\q/2+\y,\s/2-\q+\z);
			\coordinate (C) at (\x,\s/2-\q/2+\y,\s/2-\q+\z);
			\coordinate (D) at (\x,\s/2-\q/2+\y,-\s/2+\q+\z);

			% Use 'fill=none' if needed
			%\filldraw[#5] (A) -- (B) -- (C) -- (D);  % Fill before draw edges

			\draw[line width=0.2mm, color=border-red] (A) -- (B);  % Bottom line
			\draw[line width=0.2mm, color=border-red] (B) -- (C);  % Vertical front
			\draw[line width=0.2mm, color=border-red] (C) -- (D);  % Top line
			\draw[line width=0.2mm, color=border-red] (D) -- (A);  % Vertical back

			\def\d{#6} % Distance to point of pyramid
			\coordinate (E) at (\x+\d,\y,\z);
			\draw[line width=0.2mm, color=border-red] (A) -- (E);  %
			\draw[line width=0.2mm, color=border-red] (B) -- (E);  %
			\draw[line width=0.2mm, color=border-red] (C) -- (E);  %
			\draw[line width=0.2mm, color=border-red] (D) -- (E);  %

		}


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

		% INPUT
		% \networkLayer{6.0}{0}{0}{0}{color=gray!80};
		\node[canvas is zy plane at x=0,draw,fill=white] at (0,0) {\includegraphics[scale=.14]{images/bird.jpeg}};

		\pyramid{1.28}{0}{1.79}{1.43}{color=blue!50, opacity=0.5}{4};

		% First convolutional layer
 		\foreach \i in {0,...,5}
   			\networkLayer{5}{4+\i*2/10}{0}{0}{color=fill-light-blue};

		\pyramid{0.5}{5.4}{1}{1}{color=gray!10}{3};  % Last arument is depth to point


		% First maxpool layer
 		\foreach \i in {0,...,7}
   			\networkLayer{3}{8.4+\i*2/10}{0}{0}{color=fill-light-blue};

		\pyramid{0.8}{9.8}{1}{0.8}{color=gray!10}{3};


		% Second convolutional layer
 		\foreach \i in {0,...,11}
   			\networkLayer{2.5}{12.8+\i*2/10}{0}{0}{color=fill-light-blue};

		\pyramid{0.5}{15.3}{1}{1.2}{color=gray!10}{3};

		% Second maxpool layer
 		\foreach \i in {0,...,15}
   			\networkLayer{1.5}{17.8+\i*2/10}{0}{0}{color=fill-light-blue};

		\coordinate (A1) at (22.8,-3,0) ;
		\coordinate (A2) at (22.8,3,0) ;
		\coordinate (B1) at (24.8,-1,0) ;
		\coordinate (B2) at (24.8,1,0) ;
		\coordinate (C1) at (26.8,-1,0) ;
		\coordinate (C2) at (26.8,1,0) ;
		\coordinate (A11) at (20.5,-1,0) ;
		\coordinate (A12) at (20.5,0.45,0) ;

		% Reshape
		\draw[line width=1mm] (A1) -- (A2);
		\draw[line width=0.1mm] (A) -- (A1);
		\draw[line width=0.1mm] (D) -- (A2);

		% Fully connected Dense (dim of 10!)
		\draw[line width=1mm] (B1) -- (B2);
		\draw[line width=0.1mm] (A1) -- (A11);
		\draw[line width=0.1mm] (A2) -- (A12);
		\draw[line width=0.1mm] (A1) -- (B1);
		\draw[line width=0.1mm] (A1) -- (B2);
		\draw[line width=0.1mm] (A2) -- (B1);
		\draw[line width=0.1mm] (A2) -- (B2);

   		\node[Arrow]  at (25.8,0,0) {};

	   	\node[text width=3cm, align=center%, font=\fontsize{45}{1}\selectfont
     ]  at (27.5,0,0) {\Huge {\texttt{Bird}}};

	\end{tikzpicture}
	%}

\end{document}