Search

Three-dimensional (3D) Scene Understanding achieves environmental perception by extracting and analyzing point cloud data with wide applications including virtual reality, robotics, etc. Previous methods align the 2D image feature from a pre-trained CLIP model and the 3D point cloud feature for the open vocabulary scene understanding ability. We believe that existing methods have the following two deficiencies: (1) the 3D feature extraction process ignores the challenges of real scenarios, i.e., point cloud data are very sparse and even incomplete; (2) the training stage lacks direct text supervision, leading to inconsistency with the inference stage. To address the first issue, we employ a Masked Consistency training policy. Specifically, during the alignment of 3D and 2D features, we mask some 3D features to force the model to understand the entire scene using only partial 3D features. For the second issue, we generate pseudo-text labels and align them with the 3D features during the training process. In particular, we first generate a description for each 2D image belonging to the same 3D scene and then use a summarization model to fuse these descriptions into a single description of the scene. Subsequently, we align 2D-3D features and 3D-text features simultaneously during training. Massive experiments demonstrate the effectiveness of our method, outperforming state-of-the-art approaches. [ABSTRACT FROM AUTHOR]

JPEG is the international standard for still image encoding and is the most widely used compression algorithm because of its simple encoding process and low computational complexity. Recently, many methods have been developed to improve the quality of JPEG images by using deep learning. However, these methods require the use of high-performance devices since they need to perform neural network computation for decoding images. In this paper, we propose a method to generate high-quality images using deep learning without changing the decoding algorithm. The key idea is to reduce and smooth colors and gradient regions in the original images before JPEG compression. The reduction and smoothing can suppress red block noise and pseudo-contour in the compressed images. Furthermore, high-performance devices are unnecessary for decoding. The proposed method consists of two components: a color transformation network using deep learning and a pseudo-contour suppression model using signal processing. The experimental results showed that the proposed method outperforms standard JPEG in quality measurements correlated with human perception. [ABSTRACT FROM AUTHOR]

Recently, attention has surged concerning intelligent sensors using text detection. However, there are challenges in detecting small texts. To solve this problem, we propose a novel text detection CNN (convolutional neural network) architecture sensitive to text scale. We extract multi-resolution feature maps in multi-stage convolution layers that have been employed to prevent losing information and maintain the feature size. In addition, we developed the CNN considering the receptive field size to generate proposal stages. The experimental results show the importance of the receptive field size. [ABSTRACT FROM AUTHOR]