Video Experimental Relacionado
Updated: Jun 13, 2026
12:08
From Voxels to Knowledge: A Practical Guide to the Segmentation of Complex Electron Microscopy 3D-Data
Published on: August 13, 2014
25.0K
Puente entre métodos clásicos y neuronales para mejorar la segmentación en imágenes de texto matemático
Sakshi1, Chetan Sharma2, Vivek Bhardwaj3
1Amity Institute of Information Technology, Amity University, Noida, India.
Scientific reports
|December 13, 2025
Resumen
Este estudio presenta una red neuronal óptima para la segmentación de expresiones matemáticas escritas a mano, mejorando significativamente la precisión del reconocimiento. El nuevo método supera a las técnicas tradicionales, mejorando las aplicaciones de visión por computadora.
Área de la Ciencia:
- Visión por Computadora
- Procesamiento de Imágenes
- Inteligencia Artificial
Sus antecedentes:
- El reconocimiento de expresiones matemáticas se ve obstaculizado por desafíos de segmentación.
- La investigación existente prioriza el reconocimiento sobre la segmentación, particularmente en visión por computadora y procesamiento de imágenes.
- El reconocimiento de texto y expresiones matemáticas escritas a mano requiere una segmentación robusta.
Objetivo del estudio:
- Abordar el problema crítico de segmentación en el reconocimiento de expresiones matemáticas escritas a mano.
- Analizar y desarrollar una solución de segmentación óptima para expresiones matemáticas.
- Mejorar la precisión y robustez de los sistemas de reconocimiento de expresiones matemáticas.
Principales métodos:
- Exploración, clasificación y prueba de métodos de segmentación clásicos.
- Análisis comparativos de casos en diversos conjuntos de datos de expresiones matemáticas.
- Desarrollo y propuesta de un enfoque de segmentación óptimo basado en redes neuronales.
Principales resultados:
- El modelo de red neuronal propuesto logró puntuaciones competitivas de Intersection over Union (IOU) media en múltiples conjuntos de datos (CROHME, Aidapearson, HasyV).
- Las métricas de rendimiento incluyeron 79.4% (CROHME 2014), 83.5% (CROHME 2016), 81.3% (CROHME 2019), 74.6% (Aidapearson) y 79.6% (HasyV).
- El enfoque de red neuronal demostró un rendimiento superior en comparación con los métodos de segmentación tradicionales.
Conclusiones:
- La segmentación basada en redes neuronales supera eficazmente las limitaciones de los métodos clásicos.
- La solución propuesta muestra un potencial significativo para mejorar los sistemas de reconocimiento de expresiones matemáticas.
- Este trabajo avanza las técnicas de segmentación cruciales para la visión por computadora y la IA en contextos matemáticos.
Videos de Conceptos Relacionados
01:21Introduction to Scalers
Many familiar physical quantities can be specified completely by giving a single number and the appropriate unit. For example, "a class period lasts 50 min," or "the gas tank in my car holds 65 L," or "the distance between the two posts is 100 m." A physical quantity that can be specified completely in this manner is called a scalar quantity. The word "scalar" is a synonym for "number." Time, mass, distance, length, volume, temperature, and energy are some examples of scalar quantities.
Scalar...
Scalar...
01:24Convolution: Math, Graphics, and Discrete Signals
In any LTI (Linear Time-Invariant) system, the convolution of two signals is denoted using a convolution operator, assuming all initial conditions are zero. The convolution integral can be divided into two parts: the zero-input or natural response and the zero-state or forced response, with t0 indicating the initial time.
To simplify the convolution integral, it is assumed that both the input signal and impulse response are zero for negative time values. The graphical convolution process...
To simplify the convolution integral, it is assumed that both the input signal and impulse response are zero for negative time values. The graphical convolution process...
01:10Reconstruction of Signal using Interpolation
Signal processing techniques are essential for accurately converting continuous signals to digital formats and vice versa. When a continuous signal is sampled with a period T, the resulting sampled signal exhibits replicas of the original spectrum in the frequency domain, spaced at intervals equal to the sampling frequency. To handle this sampled signal, a zero-order hold method can be applied, which creates a piecewise constant signal by retaining each sample's value until the next sampling...
01:29Transformations of Functions I
A function's graph can be modified by changing its position or size without altering its overall shape. These transformations allow the graph to be moved across the coordinate plane while preserving its pattern and structure. One of the most common transformations is shifting, which repositions the graph without distorting it.When the output of a function is adjusted by adding or subtracting a constant, the graph shifts vertically. A positive value moves the graph upward, while a negative value...
01:29Transformations of Functions II
Transformations in mathematics alter the position or orientation of a function’s graph while preserving its fundamental shape. One important type of transformation is the horizontal shift, which involves modifying the input variable within a function’s equation. This operation affects where outputs occur along the horizontal axis but does not alter the function’s overall structure.A horizontal shift is achieved by replacing the input variable x with either x + c or x - c, where c is a constant.
01:20Transformations of Functions III
Transformations modify the graphical representation of a function without changing its fundamental form. One common transformation is reflection, which flips the graph across a designated axis. When the vertical coordinates of all points are multiplied by the negative one, the entire graph is mirrored over the horizontal axis. This transformation reverses the vertical orientation of peaks and troughs, akin to signal inversion in electrical systems, where a waveform is flipped, but the timing of...