Search

This paper focuses on the design of Convolution Neural Networks to visually guide an autonomous Unmanned Aerial Vehicle required to inspect power towers. The network is required to precisely segment images taken by a camera mounted on a UAV in order to allow a motion module to generate collision-free and inspection-relevant manoeuvres of the UAV along different types of towers. The images segmentation process is particularly challenging not only because of the different structures of the towers but also because of the enormous variability of the background, which can vary from the uniform blue of the sky to the multi-colour complexity of a rural, forest, or urban area. To be able to train networks that are robust enough to deal with the task variability, without incurring into a labour-intensive and costly annotation process of physical-world images, we have carried out a comparative study in which we evaluate the performances of networks trained either with synthetic images (i.e., the synthetic dataset), physical-world images (i.e., the physical-world dataset), or a combination of these two types of images (i.e., the hybrid dataset). The network used is an attention-based U-NET. The synthetic images are created using photogrammetry, to accurately model power towers, and simulated environments modelling a UAV during inspection of different power towers in different settings. Our findings reveal that the network trained on the hybrid dataset outperforms the networks trained with the synthetic and the physical-world image datasets. Most notably, the networks trained with the hybrid dataset demonstrates a superior performance on multiples evaluation metrics related to the image-segmentation task. This suggests that, the combination of synthetic and physical-world images represents the best trade-off to minimise the costs related to capturing and annotating physical-world images, and to maximise the task performances. Moreover, the results of our study demonstrate the potential of photogrammetry in creating effective training datasets to design networks to automate the precise movement of visually-guided UAVs. [ABSTRACT FROM AUTHOR]

In recent years, UNet and its derivative networks have gained widespread recognition as major methods of medical image segmentation. However, networks like UNet often struggle with Point-of-Care (POC) healthcare applications due to their high number of parameters and computational complexity. To tackle these challenges, this paper introduces an efficient network designed for medical image segmentation called MCU-Net, which leverages ConvNeXt to enhance UNet. 1) Based on ConvNeXt, MCU-Net proposes the MCU Block, which employs techniques such as large kernel convolution, depth-wise separable convolution, and an inverted bottleneck design. To ensure stable segmentation performance, it also integrates global response normalization (GRN) layers and Gaussian Error Linear Unit (GELU) activation functions. 2) Additionally, MCU-Net introduces an enhanced Multi-Scale Convolution Attention (MSCA) module after the original UNet's skip connections, emphasizing medical image features and capturing semantic insights across multiple scales. 3)The downsampling process replaces pooling layers with convolutions, and both upsampling and downsampling stages incorporate batch normalization (BN) layers to enhance model stability during training. The experimental results demonstrate that MCU-Net, with a parameter count of 2.19 million and computational complexity of 19.73 FLOPs, outperforms other segmentation models. The overall performance of MCU-Net in medical image segmentation surpasses that of other models, achieving a Dice score of 91.8% and mIoU of 84.7% on the GlaS dataset. When compared to UNet on the BUSI dataset, MCU-Net shows an improvement of 2% in Dice and 2.9% in mIoU. [ABSTRACT FROM AUTHOR]

Developing 3D Convolutional Neural Networks (CNNs) for medical image segmentation is challenging due to the limited number of available labeled medical images and computational resources. Effective design of a 3D CNN requires a better structure and appropriate hyperparameters. Manually designing a neural network and choosing appropriate hyperparameters requires extensive knowledge, a time-consuming and challenging process, resulting in complex and over-parameterized networks. Neuroevolution is an automatic and effective approach to finding an optimal network structure and hyperparameters. This paper proposes an evolutionary based method using Chameleon Swarm Algorithm (CSA) to develop 3D CNN networks for volume segmentation. We also design a search space to build optimal 3D block-based encoder–decoder structure networks for medical image segmentation. We evaluated the proposed method on two different datasets (CT Spleen and MRI Heart) and compared the results of the obtained model with state-of-the-art models. Furthermore, the comparative results show that our proposed method yields 3D CNN networks with superior performance and fewer parameters than other state-of-the-art models utilized in 3D medical image segmentation. [Display omitted] • Proposed a dynamic 3D CNN for medical image segmentation with fewer parameters. • Implemented search space with various blocks, parameters for encoder–decoder network. • Employed CSA to search the optimal network structure and training hyperparameters. • Results demonstrate the effectiveness of proposed method for 3D segmentation tasks. [ABSTRACT FROM AUTHOR]