The auction algorithm is a widely used method for task assignments. However, most existing auction algorithms yield poor performance when applied to multi-UAVs dynamic task assignment. To end this, we propose a novel hybrid “Two-Stage” auction algorithm based on the hierarchical decision mechanism and an improved objective function, which simultaneously realizes heterogeneous multi-UAVs dynamic task assignment with limited resources of each UAV and avoidance obstacle path planning. In the first stage, according to the novel proposed hierarchical decision mechanism, we select a task that is urgently needed to be performed in the task group by using the decision function and three attribute values of tasks. After the first stage, it will result in a reasonable auction sequence, instead of random auction sequence as in previous algorithms. In the second stage, by considering the coverage factor and adaptive-limitation penalty term, a novel objective function is proposed and directs related UAVs for auction. In addition, we combine the structural advantages of the centralized and distributed auction algorithm, which greatly promotes its performance in dynamic task assignment. The experimental results demonstrate that the proposed method outperforms many state-of-the-art models in efficiency and robustness.
The active contour model (ACM) is a popular approach for image segmentation. Many existing ACMs perform poorly in severe inhomogeneous images. To address this issue, a novel local and global ACM (LaG_ACM) is proposed in this paper. First, we define a global fitting image formulation that encodes the global property of an image and a global energy term using the relative entropy between the original image and the proposed global fitting image formulation. Then, a local image bias field formulation is defined to extract the local image information and to estimate the bias field. By integrating the proposed local image bias field formulation with the ACM, we specify a local energy term using the mean squared error to accommodate severe inhomogeneous images. More importantly, we define an adaptive weighting function using image entropy, which can automatically adjust the weight between the local and global energy terms according to the degree of intensity inhomogeneity. Finally, the experimental results on images with different degrees of intensity inhomogeneity validate the favorable performance of the LaG_ACM.
2 December 2023 in Expert Systems with Applications
Contrastive Language-Image Pre-training (CLIP) starts to emerge in many computer vision tasks and has achieved promising performance. However, it remains underexplored whether CLIP can be generalized to 3D hand pose estimation, as bridging text prompts with pose-aware features presents significant challenges due to the discrete nature of joint positions in 3D space. In this paper, we make one of the first attempts to propose a novel 3D hand pose estimator from monocular images, dubbed as CLIP-Hand3D, which successfully bridges the gap between text prompts and irregular detailed pose distribution. In particular, the distribution order of hand joints in various 3D space directions is derived from pose labels, forming corresponding text prompts that are subsequently encoded into text representations. Simultaneously, 21 hand joints in the 3D space are retrieved, and their spatial distribution (in x, y, and z axes) is encoded to form pose-aware features. Subsequently, we maximize semantic consistency for a pair of pose-text features following a CLIP-based contrastive learning paradigm. Furthermore, a coarse-to-fine mesh regressor is designed, which is capable of effectively querying joint-aware cues from the feature pyramid. Extensive experiments on several public hand benchmarks show that the proposed model attains a significantly faster inference speed while achieving state-of-the-art performance compared to methods utilizing the similar scale backbone. Code is available at: https://github.com/ShaoXiang23/CLIP_Hand_Demo.