CST-YOLO: A Novel Method for Blood Cell Detection Based on Improved YOLOv7 and CNN-Swin Transformer (2023-06-26T00:00:00.000000Z)

TL;DR

This paper proposes a CST-YOLO model for blood cell detection based on YOLOv7 architecture and enhances it with the CNN-Swin Transformer (CST), which is a new attempt at CNN-Transformer fusion.

Abstract

Blood cell detection is a typical small-scale object detection problem in computer vision. In this paper, we propose a CST-YOLO model for blood cell detection based on YOLOv7 architecture and enhance it with the CNN-Swin Transformer (CST), which is a new attempt at CNN-Transformer fusion. We also introduce three other useful modules: Weighted Efficient Layer Aggregation Networks (W-ELAN), Multiscale Channel Split (MCS), and Concatenate Convolutional Layers (CatConv) in our CST-YOLO to improve small-scale object detection precision. Experimental results show that the proposed CST-YOLO achieves 92.7%, 95.6%, and 91.1% mAP@0.5, respectively, on three blood cell datasets, outperforming state-of-the-art object detectors, e.g., RT-DETR, YOLOv5, and YOLOv7. Our code is available at https://github.com/mkang315/CST-YOLO.

Authors

Ming Kang

1 papers

C. Ting

1 papers

F. F. Ting

1 papers

TL;DR

Abstract

Authors

References34 items

CNN and swin-transformer based efficient model for Alzheimer's disease diagnosis with sMRI

DETRs Beat YOLOs on Real-time Object Detection

YOLOv6 v3.0: A Full-Scale Reloading

Blood Cell Count and Detection Method Based on YOLO

Designing Network Design Strategies Through Gradient Path Analysis

CST: Convolutional Swin Transformer for detecting the degree and types of plant diseases

CSwin-PNet: A CNN-Swin Transformer combined pyramid network for breast lesion segmentation in ultrasound images

YOLOv6: A Single-Stage Object Detection Framework for Industrial Applications

Complete Blood Cell Detection and Counting Based on Deep Neural Networks

YOLOv7: Trainable Bag-of-Freebies Sets New State-of-the-Art for Real-Time Object Detectors

Small Object Detection via Pixel Level Balancing With Applications to Blood Cell Detection

SwinNet: Swin Transformer Drives Edge-Aware RGB-D and RGB-T Salient Object Detection

PP-YOLOE: An evolved version of YOLO

A real time intelligent detection and counting method of cells based on YOLOv5

ISE-YOLO: Improved Squeeze-and-Excitation Attention Module based YOLO for Blood Cells Detection

YOLOX: Exceeding YOLO Series in 2021

PP-YOLOv2: A Practical Object Detector

RepVGG: Making VGG-style ConvNets Great Again

Scale Match for Tiny Person Detection

CSPNet: A New Backbone that can Enhance Learning Capability of CNN

EfficientDet: Scalable and Efficient Object Detection

Machine learning approach of automatic identification and counting of blood cells

EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks

ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design

Sigmoid-Weighted Linear Units for Neural Network Function Approximation in Reinforcement Learning

Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization

Spatial Pyramid Pooling in Deep Convolutional Networks for Visual Recognition

Yolov8 by ultralytics

Yolo by ultralytics (version 8.0.163)

Swin Transformer: Hierarchical Vision Transformer using Shifted Windows

Complete blood count (cbc) dataset

Blood cell detection dataset

We apply a Multiscale Channel Split (MCS) module to extract feature information from the input feature map with different receptive fields

We develop an adaptive feature fusion module called Weighted ELAN (W-ELAN), which can dynamically fuse the feature map data with the effective feature map and restrain the invalid feature map

Field of Study

Journal Information

Name

Volume

Venue Information

Name

Type

URL

Alternate Names