PhyTracker: An Online Tracker for Phytoplankton

TL;DR

This work introduces PhyTracker, an intelligent in situ tracking framework designed for automatic tracking of phytoplankton, highlighting key differences between phytoplankton and traditional objects, offering an effective solution for phytoplankton monitoring.

Abstract

Phytoplankton, a crucial component of aquatic ecosystems, requires efficient monitoring to understand marine ecological processes and environmental conditions. Traditional phytoplankton monitoring methods, relying on non-in situ observations, are time-consuming and resource-intensive, limiting timely analysis. To address these limitations, we introduce PhyTracker, an intelligent in situ tracking framework designed for automatic tracking of phytoplankton. PhyTracker overcomes significant challenges unique to phytoplankton monitoring, such as constrained mobility within water flow, inconspicuous appearance, and the presence of impurities. Our method incorporates three innovative modules: a Texture-enhanced Feature Extraction (TFE) module, an Attention-enhanced Temporal Association (ATA) module, and a Flow-agnostic Movement Refinement (FMR) module. These modules enhance feature capture, differentiate between phytoplankton and impurities, and refine movement characteristics, respectively. Extensive experiments on the PMOT dataset validate the superiority of PhyTracker in phytoplankton tracking, and additional tests on the MOT dataset demonstrate its general applicability, outperforming conventional tracking methods. This work highlights key differences between phytoplankton and traditional objects, offering an effective solution for phytoplankton monitoring.

Figures (11)

• Figure 1: The pedestrian dataset (on the left) and the planktonic dataset (on the right) have different characteristics.
• Figure 2: Overview of PhyTracker. PhyTracker receives the current frame picture and the previous frame picture at a time, as well as the trajectory information integrated from all previous frames, to assist in the tracking of the current frame. The yellow arrows indicate the offset in the x-coordinate calculated from the Similarity Map, while the red arrows represent the offset in the y-coordinate derived from the same map. The purple arrow represents the Memory Offset obtained from the previous frame when calculating the current frame. The golden arrow indicates the detection result generated from the previous frame's head, which contains only positional information without any class information.
• Figure 3: Each row displays the focus points of the algorithm on a single target only. The first column is the original image, the second column is DLA34+ByteTrack, and the third column is PhyTracker.
• Figure 4: We use three consecutive SIE modules(Semantic Information Extraction module) to extract semantic feature. The core of the SIE network is the SRM filter.
• Figure 5: We take the basic features of three consecutive frames as input and output the similarity information between frames.