Machine Learning Notes - A Brief Introduction to Various Target Tracking and Detection Frameworks Based on Deep Learning

1. Target tracking

        Object tracking is the task of performing an initial set of object detections, creating a unique ID for each initial detection, and then tracking each object as it moves through frames in a video, thereby maintaining the ID assignment. State-of-the-art methods involve fusing data from RGB and event-based cameras to produce more reliable object tracking. CNN-based models using only RGB images as input also work well. The most popular benchmark is OTB. There are several evaluation metrics specific to object tracking, including HOTA, MOTA, IDF1, and Track-mAP.

1. What is OTB?

        Object Tracking Benchmark (OTB) is a visual tracking benchmark widely used to evaluate the performance of visual tracking algorithms. The dataset contains a total of 100 sequences, each annotated frame-by-frame with bounding boxes and 11 challenging attributes. The OTB-2013 dataset contains 51 sequences, and the OTB-2015 dataset contains all 100 sequences of the OTB dataset.

        The following URL is the official website of OTB, which provides data sets.

Visual Tracker Benchmark (hanyang.ac.kr) http://cvlab.hanyang.ac.kr/tracker_benchmark/datasets.html         Some examples