No abstract

Abstract-The automated analysis of interacting objects or people from video has many uses, including the recognition of activities, and identification of prototypical or unusual behaviors. Existing techniques generally use temporal sequences of quantifiable real-valued features, such as object position or orientation; however, more recently, qualitative representations have been proposed. In this paper we present a novel and robust qualitative method which can be used both for classification and clustering of pair-activities. We use Qualitative Trajectory Calculus (QT C) to represent the relative motion between two objects, and encodes their interactions as a trajectory of QT C states. A key element is a general and robust means of determining the sequence similarity, which we term Normalized Weighted Sequence Alignment; we show that this is an effective metric for both recognition and clustering problems. We have evaluated our method across three different datasets, and shown that it out-performs state of the art quantitative methods, achieving an error rate of no more than 4.1% for recognition, and cluster purities higher than 90%. Our motivation originates from an interest in automated analysis of animal behaviors, and we present a comprehensive video dataset of fish behaviors (Gasterosteus aculeatus), collected from lab-based experiments.

This paper describes and evaluates a flexible, non-invasive tagging system for the automated identification and long-term monitoring of individual three-spined sticklebacks Gasterosteus aculeatus. The system is based on barcoded tags, which can be reliably and robustly detected and decoded to provide information on an individual's identity and location. Because large numbers of fish can be individually tagged, it can be used to monitor individual-and group-level dynamics within fish shoals.

No abstract