Consecutive multiscale feature learning-based image classification model

Abstract

Extracting useful features at multiple scales is a crucial task in computer vision. The emergence of deep-learning techniques and the advancements in convolutional neural networks (CNNs) have facilitated effective multiscale feature extraction that results in stable performance improvements in numerous real-life applications. However, currently available state-of-the-art methods primarily rely on a parallel multiscale feature extraction approach, and despite exhibiting competitive accuracy, the models lead to poor results in efficient computation and low generalization on small-scale images. Moreover, efficient and lightweight networks cannot appropriately learn useful features, and this causes underfitting when training with small-scale images or datasets with a limited number of samples. To address these problems, we propose a novel image classification system based on elaborate data preprocessing steps and a carefully designed CNN model architecture. Specifically, we present a consecutive multiscale feature-learning network (CMSFL-Net) that employs a consecutive feature-learning approach based on the usage of various feature maps with different receptive fields to achieve faster training/inference and higher accuracy. In the conducted experiments using six real-life image classification datasets, including small-scale, large-scale, and limited data, the CMSFL-Net exhibits an accuracy comparable with those of existing state-of-the-art efficient networks. Moreover, the proposed system outperforms them in terms of efficiency and speed and achieves the best results in accuracy-efficiency trade-off.

A Deep Learning Model for Bacterial Classification Using Big Transfer (BiT)

“…Table V compares the performance of our proposed bacteria classification models with other popular deep learning models including AlexNet [21], VGG [26], ResNet [22], DenseNet [27], SqueezeNet [28], vision transformer (ViT) [29], model soup [30] and Lion Fine-tune CNN [31]. Furthermore, we also compared the proposed method with the other two recent SOAT methods namely UzADL [32] and CMSFL [33] which used WIB-ReLU. As mentioned earlier, even though each model has its own based architecture, all of them were developed using the same parameter setting to maintain a fair basis for comparison.…”

“…Besides the aforementioned models, we also conduct a comparison of our proposed model with the most recent state-of-the-art (SOTA) image classification methods including vision transformer (ViT) [29], model soup [30], Lion Fine-tune CNN [31], UzADL [32] and CMSFL [33]. For such models, we regenerate the code and train using our bacteria dataset.…”

A Cross-Period Network for Clothing Change Person Re-Identification

“…bution of our model in the clothing change. While ResNet-50 may serve as a backbone network for fair comparison, we also integrated a better backbone CMSFL [31] to further improve the performance of our framework. The knowledge embedding configuration used in our model is based on the pretraining module described in KST-GCN [32].…”

A Lightweight Convolutional Neural Network with Hierarchical Multi-Scale Feature Fusion for Image Classification

“…However, the computational complexity of the proposed model is significantly higher compared to traditional lightweight models. This is due to the parallel processing and multiple feature extraction layers involved in the HiFuse model [37]. presented a novel image classification system called CMSFL-Net, which utilizes a consecutive multiscale feature-learning approach.…”