Tracked Instance Search
نویسندگان
چکیده
In this work we propose tracking as a generic addition to the instance search task. From video data perspective, much information that can be used is not taken into account in the traditional instance search approach. This work aims to provide insights on exploiting such existing information by means of tracking and the proper combination of the results, independently of the instance search system. We also present a study on the improvement of the system when using multiple independent instances (up to 4) of the same person. Experimental results show that our system improves substantially its performance when using tracking. Best configuration improves from mAP = 0.447 to mAP = 0.511 for a single example, and from mAP = 0.647 to mAP = 0.704 for multiple (4) given examples.
منابع مشابه
IFSB-ReliefF: A New Instance and Feature Selection Algorithm Based on ReliefF
Increasing the use of Internet and some phenomena such as sensor networks has led to an unnecessary increasing the volume of information. Though it has many benefits, it causes problems such as storage space requirements and better processors, as well as data refinement to remove unnecessary data. Data reduction methods provide ways to select useful data from a large amount of duplicate, incomp...
متن کاملThe Analysis of Vibration Characteristics and Motion Stability of the Tracked Ambulance Nonlinear Damping System
Considering the impact of the nonlinear stiffness, a 2 DOF dynamic nonlinear vibration model with cubic terms was established according to the structural feature and nonlinear behavior of the tracked ambulance. In the case of primary resonance and 1: 1 internal resonance, multiple scale method was used to obtain a first-order approximate solution for this model. Taking the parameters of the tra...
متن کاملProbabilistic Hierarchical Framework for Clustering Tracked Objects in Video Streams
Currently it is impossible to automatically achieve high level video indexing as suggested for instance by MPEG-7. Part of the problem is the identiication of moving objects in the framework of a video. It represents such a challenging problem due to the variable appearance of objects over time. This paper studies how to automatically classify tracked objects in a video based on estimated Gauss...
متن کاملNew Formulation and Solution in PCB Assembly Systems with Parallel Batch processors
This paper considers the scheduling problem of parallel batch processing machines with non-identical job size and processing time. In this paper, a new mathematical model with ready time and batch size constraints is presented to formulate the problem mathematically, in which simultaneous reduction of the makespan and earliness-tardiness is the objective function. In recent years, the nature-in...
متن کاملA Modified Discreet Particle Swarm Optimization for a Multi-level Emergency Supplies Distribution Network
Currently, the research of emergency supplies distribution and decision models mostly focus on deterministic models and exact algorithm. A few of studies have been done on the multi-level distribution network and matheuristic algorithm. In this paper, random processes theory is adopted to establish emergency supplies distribution and decision model for multi-level network. By analyzing the char...
متن کامل