Semantic segmentation is an important approach in remote sensing image analysis. However, when segmenting multiobject from remote sensing images with insufficient labeled data and imbalanced data classes, the performances of the current semantic segmentation models were often unsatisfactory. In this paper, we try to solve this problem with transfer learning and a novel deep convolutional neural network with dense connection. We designed a UNet-based deep convolutional neural network, which is called TL-DenseUNet, for the semantic segmentation of remote sensing images. The proposed TL-DenseUNet contains two subnetworks. Among them, the encoder subnetwork uses a transferring DenseNet pretrained on three-band ImageNet images to extract multilevel semantic features, and the decoder subnetwork adopts dense connection to fuse the multiscale information in each layer, which can strengthen the expressive capability of the features. We carried out comprehensive experiments on remote sensing image datasets with 11 classes of ground objects. The experimental results demonstrate that both transfer learning and dense connection are effective for the multiobject semantic segmentation of remote sensing images with insufficient labeled data and imbalanced data classes. Compared with several other state-of-the-art models, the kappa coefficient of TL-DenseUNet is improved by more than 0.0752. TL-DenseUNet achieves better performance and more accurate segmentation results than the state-of-the-art models.

Semantic segmentation methods have been successfully applied in seabed sediment detection. However, fast models like YOLO only produce rough segmentation boundaries (rectangles), while precise models like U-Net require too much time. In order to achieve fast and precise semantic segmentation results, this paper introduces a novel model called YOLO-C. It utilizes the full-resolution classification features of the semantic segmentation algorithm to generate more accurate regions of interest, enabling rapid separation of potential targets and achieving region-based partitioning and precise object boundaries. YOLO-C surpasses existing methods in terms of accuracy and detection scope. Compared to U-Net, it achieves an impressive 15.17% improvement in mean pixel accuracy (mPA). With a processing speed of 98 frames per second, YOLO-C meets the requirements of real-time detection and provides accurate size estimation through segmentation. Furthermore, it achieves a mean average precision (mAP) of 58.94% and a mean intersection over union (mIoU) of 70.36%, outperforming industry-standard algorithms such as YOLOX. Because of the good performance in both rapid processing and high precision, YOLO-C can be effectively utilized in real-time seabed exploration tasks.