Video Experimental Relacionado
Updated: Jan 15, 2026

08:06
Eye Movement Monitoring of Memory
Published on: August 15, 2010
15.2K
SAMURAI: Memoria consciente del movimiento para el seguimiento de objetos visuales sin entrenamiento con SAM 2
Resumen
SAMURAI mejora el modelo SAM 2 (Segment Anything Model 2) para el seguimiento robusto de objetos visuales. Utiliza señales de movimiento y memoria selectiva para superar los desafíos en escenas concurridas, logrando resultados de última generación sin reentrenamiento.
Área de la Ciencia:
- Visión por Computadora
- Inteligencia Artificial
- Aprendizaje Automático
Sus antecedentes:
- El Segment Anything Model 2 (SAM 2) se destaca en la segmentación de objetos, pero tiene dificultades con el seguimiento de objetos visuales, especialmente en escenarios concurridos u ocluidos.
- El mecanismo de memoria fija de SAM 2 acumula errores durante las oclusiones, lo que lleva a un seguimiento inexacto y a la deriva de la identidad.
- Los métodos existentes a menudo requieren reentrenamiento o ajuste extensos para adaptar los modelos de segmentación a tareas de seguimiento.
Objetivo del estudio:
- Presentar SAMURAI, una adaptación mejorada de SAM 2 diseñada para el seguimiento robusto de objetos visuales.
- Abordar las limitaciones de SAM 2 en el manejo de escenarios de seguimiento complejos como oclusiones y escenas concurridas.
- Desarrollar un método de seguimiento sin entrenamiento que aproveche las señales de movimiento temporal y una estrategia optimizada de selección de memoria.
Principales métodos:
- SAMURAI integra señales de movimiento temporal con una novedosa estrategia de selección de memoria consciente del movimiento.
- El modelo predice el movimiento del objeto y refina la selección de máscaras dinámicamente.
- No se requiere reentrenamiento ni ajuste del modelo SAM 2 base.
Principales resultados:
- SAMURAI demuestra un sólido rendimiento sin entrenamiento en múltiples conjuntos de datos de referencia de VOT.
- Logró resultados de vanguardia en los puntos de referencia LaSOText, GOT-10k y TrackingNet.
- Ofreció un rendimiento competitivo en los puntos de referencia LaSOT, VOT2020-ST, VOT2022-ST y SA-V.
Conclusiones:
- SAMURAI ofrece una solución robusta y precisa para el seguimiento de objetos visuales, superando las limitaciones de SAM 2.
- La estrategia de selección de memoria consciente del movimiento mejora la precisión del seguimiento en entornos dinámicos complejos.
- SAMURAI muestra un potencial significativo para aplicaciones del mundo real que requieren un seguimiento de objetos confiable.
Videos de Conceptos Relacionados
Absolute Motion Analysis- General Plane Motion
531
Visualize a drone, with its propellers spinning rapidly, hovering mid-air. The fascinating movements and operations of this drone can be comprehended by applying the principle of general plane motion.
As the drone's propellers rotate, an upward force is generated that counteracts the force of gravity, enabling the drone to lift off from the ground. This initial movement of the drone is along a straight path, representing a form of translational motion. In this phase, every point on the...
As the drone's propellers rotate, an upward force is generated that counteracts the force of gravity, enabling the drone to lift off from the ground. This initial movement of the drone is along a straight path, representing a form of translational motion. In this phase, every point on the...
531
Relative Motion Analysis using Rotating Axes
876
Consider a component AB undergoing a linear motion. Along with a linear motion, point B also rotates around point A. To comprehend this complex movement, position vectors for both points A and B are established using a stationary reference frame.
However, to express the relative position of point B relative to point A, an additional frame of reference, denoted as x'y', is necessary. This additional frame not only translates but also rotates relative to the fixed frame, making it...
However, to express the relative position of point B relative to point A, an additional frame of reference, denoted as x'y', is necessary. This additional frame not only translates but also rotates relative to the fixed frame, making it...
876
Relative Motion Analysis - Velocity
690
A stroke engine has a slider-crank mechanism that converts rotational motion from the crank into linear motion of the slider or vice versa. This mechanism consists of three main parts: the crank, the connecting rod, and the slider.
When an external force is exerted, it sets the crank into a rotational movement. This, in turn, instigates the motion of the connecting rod, leading to what is referred to as a general plane motion. This process involves two key points - point A on the connecting rod...
When an external force is exerted, it sets the crank into a rotational movement. This, in turn, instigates the motion of the connecting rod, leading to what is referred to as a general plane motion. This process involves two key points - point A on the connecting rod...
690
Relative Motion Analysis - Acceleration
810
A slider-crank mechanism converts rotational motion from the crank into linear motion of the slider or vice versa. This mechanism consists of three main parts: the crank, the connecting rod, and the slider. The movement of the slider-crank is an example of general plane motion as the fluctuating angle between the crank and the connecting rod. Consider a segment AB where point A is at the end of the slider and point B is on the diametrically opposite end to point A, on a crack. The variance in...
810
Relative Motion Analysis using Rotating Axes-Problem Solving
695
Consider a crane whose telescopic boom rotates with an angular velocity of 0.04 rad/s and angular acceleration of 0.02 rad/s2. Along with the rotation, the boom also extends linearly with a uniform speed of 5 m/s. The extension of the boom is measured at point D, which is measured with respect to the fixed point C on the other end of the boom. For the given instant, the distance between points C and D is 60 meters.
Here, in order to determine the magnitude of velocity and acceleration for point...
Here, in order to determine the magnitude of velocity and acceleration for point...
695
Observational Learning
824
Albert Bandura's observational learning, also known as imitation or modeling, occurs when a person observes and imitates another's behavior. It is a quicker process than operant conditioning. A well-known example is the Bobo doll study, where children who saw an adult acting aggressively towards the doll were more likely to act aggressively when left alone, compared to those who observed a nonaggressive adult. Many psychologists view observational learning as a form of latent learning...
824

