YOLOv5s-SA: Light-Weighted and Improved YOLOv5s for Sperm Detection

Zhu, Ronghua; Cui, Yansong; Huang, Jianming; Hou, Enyu; Zhao, Jiayu; Zhou, Zhen; Li, Hao

doi:10.3390/diagnostics13061100

Cited by 12 publications

(8 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the future, we will also compare this proposed SqueezeNet-Light architecture with other recently developed lightweight [49,50] architectures to confirm the generalizability of the network.…”

Section: Discussionmentioning

confidence: 98%

Assist-Dermo: A Lightweight Separable Vision Transformer Model for Multiclass Skin Lesion Classification

Abbas,

Daadaa,

Rashid

et al. 2023

Diagnostics

View full text Add to dashboard Cite

A dermatologist-like automatic classification system is developed in this paper to recognize nine different classes of pigmented skin lesions (PSLs), using a separable vision transformer (SVT) technique to assist clinical experts in early skin cancer detection. In the past, researchers have developed a few systems to recognize nine classes of PSLs. However, they often require enormous computations to achieve high performance, which is burdensome to deploy on resource-constrained devices. In this paper, a new approach to designing SVT architecture is developed based on SqueezeNet and depthwise separable CNN models. The primary goal is to find a deep learning architecture with few parameters that has comparable accuracy to state-of-the-art (SOTA) architectures. This paper modifies the SqueezeNet design for improved runtime performance by utilizing depthwise separable convolutions rather than simple conventional units. To develop this Assist-Dermo system, a data augmentation technique is applied to control the PSL imbalance problem. Next, a pre-processing step is integrated to select the most dominant region and then enhance the lesion patterns in a perceptual-oriented color space. Afterwards, the Assist-Dermo system is designed to improve efficacy and performance with several layers and multiple filter sizes but fewer filters and parameters. For the training and evaluation of Assist-Dermo models, a set of PSL images is collected from different online data sources such as Ph2, ISBI-2017, HAM10000, and ISIC to recognize nine classes of PSLs. On the chosen dataset, it achieves an accuracy (ACC) of 95.6%, a sensitivity (SE) of 96.7%, a specificity (SP) of 95%, and an area under the curve (AUC) of 0.95. The experimental results show that the suggested Assist-Dermo technique outperformed SOTA algorithms when recognizing nine classes of PSLs. The Assist-Dermo system performed better than other competitive systems and can support dermatologists in the diagnosis of a wide variety of PSLs through dermoscopy. The Assist-Dermo model code is freely available on GitHub for the scientific community.

show abstract

Section: Discussionmentioning

confidence: 98%

Assist-Dermo: A Lightweight Separable Vision Transformer Model for Multiclass Skin Lesion Classification

Abbas,

Daadaa,

Rashid

et al. 2023

Diagnostics

View full text Add to dashboard Cite

show abstract

“…These preprocessing tasks aim to perform encoding checks, remove non-alphanumeric characters, clean digits, normalize whitespace, convert case, tokenize, and stem words. Additionally, we utilized three machine learning methods from the sklearn library-Support Vector Machine (SVM) [35], Logistic Regression (LR) [36], and Random Forest (RF) [37]-for comparative analysis of algorithm performance.…”

Section: Machine Learning Methodsmentioning

confidence: 99%

RDscan: Extracting RNA-disease relationship from literature Using pre-training model

zhang,

Yang,

Ren

et al. 2024

Preprint

View full text Add to dashboard Cite

With the rapid advancements in molecular biology and genomics, a multitude of connections between RNA and diseases has been unveiled, making the efficient and accurate extraction of RNA-disease relationships (RD relationships) from extensive biomedical literature crucial for advancing research in this field. This study introduces RDscan, a novel text mining method developed based on the pre-training and fine-tuning strategy, aimed at automatically extracting RD-related information from a vast corpus of literature using pre-trained biomedical large language models. Initially, we constructed a dedicated RD corpus, comprising 2,082 positive and 2,000 negative statements, alongside an independent test dataset for training and evaluating RDscan. Subsequently, by fine-tuning the Bioformer and BioBERT pre-trained models, RDscan demonstrated exceptional performance in text classification and named entity recognition (NER) tasks. In 5-fold cross-validation, RDscan significantly outperformed traditional machine learning methods. In summary, RDscan represents the first text mining tool specifically designed for RD relationship extraction, and is freely available at https://github.com/ZhangCellab/RDScaning.

show abstract

“…Li et al [38] proposed an MTC-YOLOv5 algorithm for the detection of cucumber plant diseases, which added Coordinate attention (CA) to reduce background interference, combined a Multi-scale (MS) training to improve the detection accuracy of small targets, and experimental results showed that the algorithm had high detection accuracy and speed. Zhu et al developed a faster and more accurate SE-YOLOV5 model [39] proposed an improved YOLOv5 model for fast sperm detection, which added a Shuffle attention (SA) mechanism to enhance the detection performance of sperm and used DWConv to enhance the speed of convergence. Experiments showed that the model effectively reduced sperm leakage and improved detection accuracy.…”

Section: Related Workmentioning

confidence: 99%

Intelligent detection method of microparticle virus in silkworm based on YOLOv8 improved algorithm

Zhang,

Su,

Wang

et al. 2024

J Supercomput

View full text Add to dashboard Cite

The presence of microparticle viruses significantly impacts the quality of silkworm seeds for domestic sericulture, making their exclusion from detection in silkworm seed production crucial. Traditional methods for detecting microparticle viruses in silkworms, such as manual microscopic observation, molecular biology, and immunological approaches, are cumbersome and unable to achieve intelligent, batch real-time detection. To address this challenge, we employ the YOLOv8 algorithm in this paper. Firstly, NAM attention is introduced in the original algorithm’s Backbone component, allowing the model to extract more generic feature information. Secondly, ODConv replaces Conv in the Head component of the original algorithm, enhancing the model’s ability to identify microparticle viruses. Finally, NWD-LOSS modifies the CIoU loss of the original algorithm to obtain a more accurate prediction box. Experimental results demonstrate that the NN-YOLOv8 model outperforms mainstream detection algorithms in detecting silkworm microparticle diseases. With an average detection time of 22.6 milliseconds per image, the model shows promising prospects for future applications. This model improvement enhances detection efficiency and reduces human resource costs, effectively realizing detection intelligence.

show abstract

YOLOv5s-SA: Light-Weighted and Improved YOLOv5s for Sperm Detection

Cited by 12 publications

References 37 publications

Assist-Dermo: A Lightweight Separable Vision Transformer Model for Multiclass Skin Lesion Classification

Assist-Dermo: A Lightweight Separable Vision Transformer Model for Multiclass Skin Lesion Classification

RDscan: Extracting RNA-disease relationship from literature Using pre-training model

Intelligent detection method of microparticle virus in silkworm based on YOLOv8 improved algorithm

Contact Info

Product

Resources

About