31 August 2023 in Expert Systems with Applications
11 May 2023 in IEEE Transactions on Multimedia
In this work, we introduce FARP-Net, an adaptive local-global feature aggregation and relation-aware proposal network for high-quality 3D object detection from pure point clouds. Our key insight is that learning adaptive local-global feature aggregation from an irregular yet sparse point cloud and generating superb proposals are both pivotal for detection. Technically, we propose a novel local-global feature aggregation layer (LGFAL) that fully exploits the complementary correlation between local features and global features, and fuses their strengths adaptively via an attention-based fusion module. Furthermore, we incorporate a lightweight feature affine module (LFAM) into LGFAL to map the local features into a normal distribution, thus acquiring fine-grained features of each local region in a weight-sharing manner. During object proposal generation, we propose a weighted relation-aware proposal module (WRPM) that uses an objectness-aware formalism to weigh the relation importance among object candidates for a clear and principal context, thereby facilitating the generation of high-quality proposals. The WRPM challenges the traditional practice of extracting contextual information among all object candidates, which is inefficient as object candidates are always noisy and redundant. Experimentally, FARP-Net delivers superior performance on two widely used benchmarks with fewer parameters, 64.0% mAP@0.25 on the SUN RGB-D dataset and 70.9% mAP@0.25 on the ScanNet V2 dataset. We further validate that the proposed LGFAL and WRPM can be integrated into both indoor and outdoor detectors to boost performance. The code is available at https://github.com/XT-1997/FARP-Net .
31 January 2022 in IEEE Robotics and Automation Letters
As scene coordinate regression (SCoRe) methods become prevailing in the area of visual camera localization, the issue of repetitive or sparse texture scenes continues to be a concern. Specifically, they will suffer from performance degeneration due to ambiguous patterns caused by visual similarity. In this work, we propose a novel network for camera localization through a single RGB image, with our key insight that taking only high-level feature maps as input can be difficult for the network to accurately model the regression problem due to ambiguous patterns and utilizing the rich spatial details in low-level feature maps can tackle this issue. The core components of the network are 1) Deep Feature Aggregation Module (DFAM), which eliminates the difference among the different level feature representations and fuses multi-level context information; 2) CoordConv Scheme, which further improves the discrimination of features in repetitive or sparse texture areas of the image; 3) Deep Supervision, which endows low-level feature maps with direct supervision from the ground truth to improve the accuracy of camera localization; 4) Uncertainty Modeling, which quantifies the prediction errors stemming from the intrinsic noise in the data. Moreover, to maximize the power of DFAM, we embed channel attention modules into it to prune redundant and noisy features, through which we can refine the different level feature maps. Our network is designed to be lightweight and efficient, and the proposed DFAM can be integrated into general SCoRe-based networks. Comprehensive experiments demonstrate the effectiveness of DFAM and the superiority of our network over the state-of-the-art methods on two benchmarks.
24 March 2023 in IEEE Robotics and Automation Letters
Visual localization, a vital component of many visual applications, has been tackled by scene coordinates regression (SCoRe) methods that leverage neural networks to predict scene coordinates, followed by a PnP algorithm to recover camera pose. However, these methods do not consider the relationship between image patches, known as relative features or affinity information, which is instrumental for network to perform complete scene parsing. Besides, owing to the visual similarity between image patches, these methods are weak in extracting reliable absolute features that represent the context information of the image patches, resulting in inferior localization performance. In response, we propose EAAINet that is based on classical SCoRe approaches and consists of two novel modules: the Global Affinity Aggregation Module (GAAM) and the Element-wise Attention Module (EAM). Specifically, GAAM employs an interval sampling strategy to sample image patches to construct sparse graph neural networks (GNNs), from which global affinity information between image patches is retrieved, hence ensuring precise scene parsing. EAM integrates multi-level features to generate reliable absolute features to regress accurate scene coordinates, with the key insight that the structure information is essential to differentiate similar image patches and the semantic information assists in modeling regression problems. Technically, EAM predicts element-wise soft attention masks to reconcile multi-level feature maps, enabling efficient feature fusion. Positional encoding and uncertainty modeling are also employed to enhance visual localization performance. Experimental results show that EAAINet significantly outperforms the state-of-the-arts on multiple benchmarks with faster speed and less model parameters.
28 January 2023 in International Journal of Mechanical Sciences