TOD-CNN: An Effective Convolutional Neural Network for Tiny Object Detection in Sperm Videos

TL;DR

This paper introduces TOD-CNN, a convolutional neural network designed specifically for detecting tiny objects like sperms in microscopic videos, achieving high accuracy and aiding sperm quality analysis.

Contribution

The paper presents a novel CNN model tailored for tiny object detection in microscopic videos, along with a high-quality dataset and a GUI for practical application.

Findings

Achieved 85.60% AP50 in real-time sperm detection

Developed a dataset with over 278,000 annotated objects

Demonstrated the model's effectiveness in sperm quality evaluation

Abstract

The detection of tiny objects in microscopic videos is a problematic point, especially in large-scale experiments. For tiny objects (such as sperms) in microscopic videos, current detection methods face challenges in fuzzy, irregular, and precise positioning of objects. In contrast, we present a convolutional neural network for tiny object detection (TOD-CNN) with an underlying data set of high-quality sperm microscopic videos (111 videos, $>$ 278,000 annotated objects), and a graphical user interface (GUI) is designed to employ and test the proposed model effectively. TOD-CNN is highly accurate, achieving $85.60\%$ AP$_{50}$ in the task of real-time sperm detection in microscopic videos. To demonstrate the importance of sperm detection technology in sperm quality analysis, we carry out relevant sperm quality evaluation metrics and compare them with the diagnosis results from medical doctors.