Vision-language models (VLMs) have shown promise in 2D medical image analysis, but extending them to 3D remains challenging due to the high computational demands of volumetric data and the difficulty of aligning 3D spatial features with clinical text. We present Med3DVLM, a 3D VLM designed to address these challenges through three key innovations: (1) DCFormer, an efficient encoder that uses decomposed 3D convolutions to capture fine-grained spatial features at scale; (2) SigLIP, a contrastive learning strategy with pairwise sigmoid loss that improves image-text alignment without relying on large negative batches; and (3) a dual-stream MLP-Mixer projector that fuses low- and high-level image features with text embeddings for richer multi-modal representations. We evaluated our model on the M3D dataset, which includes radiology reports and VQA data for 120,084 3D medical images. The results show that Med3DVLM achieves superior performance on multiple benchmarks. For image-text retrieval, it reaches 61.00% R@1 on 2,000 samples, significantly outperforming the current state-of-the-art M3D-LaMed model (19.10%). For report generation, it achieves a METEOR score of 36.42% (vs. 14.38%). In open-ended visual question answering (VQA), it scores 36.76% METEOR (vs. 33.58%), and in closed-ended VQA, it achieves 79.95% accuracy (vs. 75.78%). These results demonstrate Med3DVLM's ability to bridge the gap between 3D imaging and language, enabling scalable, multi-task reasoning across clinical applications. Our code is publicly available at https://github.com/mirthAI/Med3DVLM.

Accurate semantic analysis of LiDAR point clouds enables the interaction between intelligent vehicles and the real environment. This paper proposes a hybrid 2D and 3D Hough Net by combining 3D global Hough features and 2D local Hough features with a classification deep learning network. Firstly, the 3D object point clouds are mapped into the 3D Hough space to extract the global Hough features. The generated global Hough features are input into the 3D convolutional neural network for training global features. Furthermore, a multi-scale critical point sampling method is designed to extract critical points in the 2D views projected from the point clouds to reduce the computation of redundant points. To extract local features, a grid-based dynamic nearest neighbors algorithm is designed by searching the neighbors of the critical points. Finally, the two networks are connected to the full connection layer, which is input into fully connected layers for object classification.

This paper presents a novel pedestrian recognition method based on Jetson Nano, leveraging its advantages as an edge computing device with low latency and high privacy. Traditional methods for pedestrian recognition often require substantial computing resources and a strong network connection, making them unsuitable for edge devices. Thus, this paper proposes an edge computing method specifically designed for Jetson Nano, utilizing CUDA and TensorRT to achieve higher performance. This method utilizes a convolutional neural network to extract image features, which are then activated through Reshape, Transpose, and Sigmoid functions. The Slice operation is employed to extract objects within areas with relatively high probabilities. The position of the object is then adjusted using mathematical operations such as Mul, Sub, Add, Pow, etc., resulting in more accurate detection results. This paper demonstrates the effectiveness of using Jetson Nano as an edge computing device for pedestrian recognition and the potential of CUDA and TensorRT for optimizing performance on edge devices. The proposed method has broad applications in various fields, such as surveillance systems, autonomous vehicles, and robotics, where real-time pedestrian recognition tasks are essential.

The existing works have issues about the angular resolution limitations of the traditional mechanical LiDAR sensor. In such instance, it will fail to gain a complete depiction of the surroundings when gathering point cloud data. Therefore, we will employ a system to extracting feature points from intersecting planes in environment to achieve reconstruction in 3D environment. The system is combined with the RANSAC algorithm and 3A-LSM to better detect the intersecting planes in the scanned point cloud environment and fit the plane equations. The extracted intersection points are used as the feature points of the current frame. The virtual angle points are aligned between frames. The frames addresses the issues of the lack of an initial value in ICP alignment algorithm and poor efficiency of alignment perform in large- scale point clouds. Eventually, the outcomes of the experiments demonstrate that the system successfully completes the global three-dimensional reconstruction.