While diffusion models have demonstrated remarkable potential in generative tasks, their application to discriminative visual perception remains challenging due to high computational costs, heavy reliance on ground-truth image-text pairs, and non-deterministic outputs. To address these limitations, we propose MF-VPD, a novel Vision Perception Diffusion Model based on Multi-Modal Feature Fusion. First, to overcome the scarcity of labeled image-text pairs, we introduce a Training Data Generator that leverages BLIP to synthesize descriptive captions. Second, we design a Multimodal Feature Fusion (MFF) module that integrates CLIP-based textual embeddings with ViT-based visual features using a state-space model strategy. This integration ensures the diffusion backbone receives comprehensive, complementary guidance, mitigating the inadequacy of unimodal text prompts. Furthermore, we propose a lightweight Visual Perception Adapter (VPA) to fine-tune the network while keeping the backbone frozen. Extensive experiments demonstrate MF-VPD’s superiority across depth estimation, semantic segmentation, and salient object detection. Notably, MF-VPD achieves an 85.17% mIoU on Cityscapes and a 0.959 S-measure on DUTS, while reducing error metrics by 6-8% on NYU Depth V2 and KITTI. Crucially, compared to the VPD baseline, our adapter strategy reduces model parameters by 76.7% and accelerates inference speed by approximately 48% on depth estimation tasks. The code will be made available at https://github.com/zhx441/MF-VPD/.

In this work, we propose a fast convergence track net, or FC-TrackNet, based on a synthetic data-driven approach to maintaining long-term 6D pose tracking. Comparison experiments are performed on two different datasets, The results demonstrate that our approach can achieve a consistent tracking frequency of 90.9 Hz as well as higher accuracy than the state-of-the art approaches.

In this paper, we propose BRAFT, an improved deep network architecture based on the Recurrent All-Pairs Field Transforms (RAFT) for optical flow estimation. BRAFT extracts features for each pixel. According to the characteristics of optical flow, the dense visual similarity is calculated based on the strength-weakness correlation in blocks to establish a more precise 4D correlation volume. Using a single dataset for training, the proposed method achieves better results than the original RAFT. We consider the end-point-error as the performance measure. The results show that the proposed method is 1.7$\%$lower than RAFT on the KITTI 2015 benchmark, and 1.7$\%$lower than MPI Sintel (final). In addition, the errors in BRAFT are mostly small. Therefore, the proposed method is better for performing the edge extraction of moving objects.

Ocean and Land Color Imager (OLCI) bio-optical products are validated using in situ data in the South China sea (SCS) and the East China sea (ECS) acquired during September and October 2018. Results show that reflectance and Kd490 are both reasonably estimated, whose APD are around 20% for both FR and RR data. However, chlorophyll concentration and TSM concentration cannot be estimated correctly. This indicates that atmospheric works quite well but the bio-optical model does not work at all. Therefore, local or regional bio-optical algorithms must be developed.

Edge is one of the most important image features, and it is the base of follow-up measurement and shape registration. In order to get better edge information, an edge detection method of block distance is proposed in this paper. Firstly, by calculating distance between the pixels within sub-block regions, the graph of Gaussian weighted distance is obtained. Comparing with the original figure, it not only can better highlight the edge contour, but also can be uniformed background gray. Secondly, in order to reduce the complexity of the algorithm, edge detection method of block distance combining with summed area table is raised. Finally, implementation process of this algorithm is raised. Experimental results show that this method can achieve high speed of edge detection with higher accuracy in extracting edge and corner information of image.