This is a workinprogress demonstrator of a Mathematical Expression Recognizer (Last update: September, 2016).
The recognition engine is based on SESHAT, an opensource system for recognizing handwritten math expressions.
At present, the system is trained to recognize more than 100 symbols:
az
AZ
09
<
>
≤
≥
=
≠

+
±
×
!

.
,
...
(
)
[
]
{
}
′
→
∃
∀
∈
∫
√
∞
α
β
Δ
γ
λ
μ
φ
π
σ
Σ
∏
θ
lim
log
cos
sin
tan
The system expects complete equations, hence, an expression like "\(2+2=\)" would not be recognized because an equal sign requires a righthand operator (e.g. "\(2+2=4\)").
Most of math expressions are accepted, containing subscripts (\(a_1\)), superscripts (\(x^2\)), fractions (\(\displaystyle\frac{a+b}{c}\)), square roots (\(\sqrt{\pi^2}\) or \(\sqrt[3]{4y}\)), integral (\(\displaystyle\int_a^b x\)), summation (\(\displaystyle\sum_{i=0}^n x+i\)), limits (\(\displaystyle\lim_{x \rightarrow \infty}\frac{1}{x}\)), functions (\(\sin^2 \alpha\)), ...
On the other hand, matrices \(\displaystyle\left(\begin{matrix} 1 & 2 \\ 3 & 4 \end{matrix}\right)\), left scripts (\({}_nC_m\)) or piecewise functions (\( f(n) = \left\{ \begin{array}{l l} n & \quad n \geq 0\\ 1n & \quad n < 0 \end{array} \right. \)) are not recognized yet.
It is expected to increase the number of symbols and accepted expressions in the shortterm, as well as improving the recognition system and the expression accepted.