Applying large-scale pre-trained visual models like CLIP to few-shot action recognition tasks can benefit performance and efficiency. Utilizing the "pre-training, fine-tuning" paradigm makes it possible to avoid training a network from scratch, which can be time-consuming and resource-intensive. However, this method has two drawbacks. First, limited labeled samples for few-shot action recognition necessitate minimizing the number of tunable parameters to mitigate over-fitting, also leading to inadequate fine-tuning that increases resource consumption and may disrupt the generalized representation of models. Second, the video's extra-temporal dimension challenges few-shot recognition's effective temporal modeling, while pre-trained visual models are usually image models. This paper proposes a novel method called Multimodal Adaptation of CLIP (MA-CLIP) to address these issues. It adapts CLIP for few-shot action recognition by adding lightweight adapters, which can minimize the number of learnable parameters and enable the model to transfer across different tasks quickly. The adapters we design can combine information from video-text multimodal sources for task-oriented spatiotemporal modeling, which is fast, efficient, and has low training costs. Additionally, based on the attention mechanism, we design a text-guided prototype construction module that can fully utilize video-text information to enhance the representation of video prototypes. Our MA-CLIP is plug-and-play, which can be used in any different few-shot action recognition temporal alignment metric.
Neural network quantization is a very promising solution in the field of model compression, but its resulting accuracy highly depends on a training/fine-tuning process and requires the original data. This not only brings heavy computation and time costs but also is not conducive to privacy and sensitive information protection. Therefore, a few recent works are starting to focus on data-free quantization. However, data-free quantization does not perform well while dealing with ultra-low precision quantization. Although researchers utilize generative methods of synthetic data to address this problem partially, data synthesis needs to take a lot of computation and time. In this paper, we propose a data-free mixed-precision compensation (DF-MPC) method to recover the performance of an ultra-low precision quantized model without any data and fine-tuning process. By assuming the quantized error caused by a low-precision quantized layer can be restored via the reconstruction of a high-precision quantized layer, we mathematically formulate the reconstruction loss between the pre-trained full-precision model and its layer-wise mixed-precision quantized model. Based on our formulation, we theoretically deduce the closed-form solution by minimizing the reconstruction loss of the feature maps. Since DF-MPC does not require any original/synthetic data, it is a more efficient method to approximate the full-precision model. Experimentally, our DF-MPC is able to achieve higher accuracy for an ultra-low precision quantized model compared to the recent methods without any data and fine-tuning process.
Semantic scene completion (SSC) jointly predicts the semantics and geometry of the entire 3D scene, which plays an essential role in 3D scene understanding for autonomous driving systems. SSC has achieved rapid progress with the help of semantic context in segmentation. However, how to effectively exploit the relationships between the semantic context in semantic segmentation and geometric structure in scene completion remains under exploration. In this paper, we propose to solve outdoor SSC from the perspective of representation separation and BEV fusion. Specifically, we present the network, named SSC-RS, which uses separate branches with deep supervision to explicitly disentangle the learning procedure of the semantic and geometric representations. And a BEV fusion network equipped with the proposed Adaptive Representation Fusion (ARF) module is presented to aggregate the multi-scale features effectively and efficiently. Due to the low computational burden and powerful representation ability, our model has good generality while running in real-time. Extensive experiments on SemanticKITTI demonstrate our SSC-RS achieves state-of-the-art performance.
Reliable LiDAR panoptic segmentation (LPS), including both semantic and instance segmentation, is vital for many robotic applications, such as autonomous driving. This work proposes a new LPS framework named PANet to eliminate the dependency on the offset branch and improve the performance on large objects, which are always over-segmented by clustering algorithms. Firstly, we propose a non-learning Sparse Instance Proposal (SIP) module with the ``sampling-shifting-grouping" scheme to directly group thing points into instances from the raw point cloud efficiently. More specifically, balanced point sampling is introduced to generate sparse seed points with more uniform point distribution over the distance range. And a shift module, termed bubble shifting, is proposed to shrink the seed points to the clustered centers. Then we utilize the connected component label algorithm to generate instance proposals. Furthermore, an instance aggregation module is devised to integrate potentially fragmented instances, improving the performance of the SIP module on large objects. Extensive experiments show that PANet achieves state-of-the-art performance among published works on the SemanticKITII validation and nuScenes validation for the panoptic segmentation task.
3D LiDAR-based single object tracking (SOT) has gained increasing attention as it plays a crucial role in 3D applications such as autonomous driving. The central problem is how to learn a target-aware representation from the sparse and incomplete point clouds. In this paper, we propose a novel Correlation Pyramid Network (CorpNet) with a unified encoder and a motion-factorized decoder. Specifically, the encoder introduces multi-level self attentions and cross attentions in its main branch to enrich the template and search region features and realize their fusion and interaction, respectively. Additionally, considering the sparsity characteristics of the point clouds, we design a lateral correlation pyramid structure for the encoder to keep as many points as possible by integrating hierarchical correlated features. The output features of the search region from the encoder can be directly fed into the decoder for predicting target locations without any extra matcher. Moreover, in the decoder of CorpNet, we design a motion-factorized head to explicitly learn the different movement patterns of the up axis and the x-y plane together. Extensive experiments on two commonly-used datasets show our CorpNet achieves state-of-the-art results while running in real-time.
Recently, neural implicit surfaces have become popular for multi-view reconstruction. To facilitate practical applications like scene editing and manipulation, some works extend the framework with semantic masks input for the object-compositional reconstruction rather than the holistic perspective. Though achieving plausible disentanglement, the performance drops significantly when processing the indoor scenes where objects are usually partially observed. We propose RICO to address this by regularizing the unobservable regions for indoor compositional reconstruction. Our key idea is to first regularize the smoothness of the occluded background, which then in turn guides the foreground object reconstruction in unobservable regions based on the object-background relationship. Particularly, we regularize the geometry smoothness of occluded background patches. With the improved background surface, the signed distance function and the reversedly rendered depth of objects can be optimized to bound them within the background range. Extensive experiments show our method outperforms other methods on synthetic and real-world indoor scenes and prove the effectiveness of proposed regularizations.
Localization and mapping with heterogeneous multi-sensor fusion have been prevalent in recent years. To adequately fuse multi-modal sensor measurements received at different time instants and different frequencies, we estimate the continuous-time trajectory by fixed-lag smoothing within a factor-graph optimization framework. With the continuous-time formulation, we can query poses at any time instants corresponding to the sensor measurements. To bound the computation complexity of the continuous-time fixed-lag smoother, we maintain temporal and keyframe sliding windows with constant size, and probabilistically marginalize out control points of the trajectory and other states, which allows preserving prior information for future sliding-window optimization. Based on continuous-time fixed-lag smoothing, we design tightly-coupled multi-modal SLAM algorithms with a variety of sensor combinations, like the LiDAR-inertial and LiDAR-inertial-camera SLAM systems, in which online timeoffset calibration is also naturally supported. More importantly, benefiting from the marginalization and our derived analytical Jacobians for optimization, the proposed continuous-time SLAM systems can achieve real-time performance regardless of the high complexity of continuous-time formulation. The proposed multi-modal SLAM systems have been widely evaluated on three public datasets and self-collect datasets. The results demonstrate that the proposed continuous-time SLAM systems can achieve high-accuracy pose estimations and outperform existing state-of-the-art methods. To benefit the research community, we will open source our code at ~\url{https://github.com/APRIL-ZJU/clic}.
In this letter, a novel method for change detection is proposed using neighborhood structure correlation. Because structure features are insensitive to the intensity differences between bi-temporal images, we perform the correlation analysis on structure features rather than intensity information. First, we extract the structure feature maps by using multi-orientated gradient information. Then, the structure feature maps are used to obtain the Neighborhood Structural Correlation Image (NSCI), which can represent the context structure information. In addition, we introduce a measure named matching error which can be used to improve neighborhood information. Subsequently, a change detection model based on the random forest is constructed. The NSCI feature and matching error are used as the model inputs for training and prediction. Finally, the decision tree voting is used to produce the change detection result. To evaluate the performance of the proposed method, it was compared with three state-of-the-art change detection methods. The experimental results on two datasets demonstrated the effectiveness and robustness of the proposed method.
Deep learning-based change detection using remote sensing images has received increasing attention in recent years. However, how to effectively extract and fuse the deep features of bi-temporal images to improve the accuracy of change detection is still a challenge. To address that, a novel adjacent-level feature fusion network with 3D convolution (named AFCF3D-Net) is proposed in this article. First, through the inner fusion property of 3D convolution, we design a new feature fusion way that can simultaneously extract and fuse the feature information from bi-temporal images. Then, in order to bridge the semantic gap between low-level features and high-level features, we propose an adjacent-level feature cross-fusion (AFCF) module to aggregate complementary feature information between the adjacent-levels. Furthermore, the densely skip connection strategy is introduced to improve the capability of pixel-wise prediction and compactness of changed objects in the results. Finally, the proposed AFCF3D-Net has been validated on the three challenging remote sensing change detection datasets: Wuhan building dataset (WHU-CD), LEVIR building dataset (LEVIR-CD), and Sun Yat-Sen University (SYSU-CD). The results of quantitative analysis and qualitative comparison demonstrate that the proposed AFCF3D-Net achieves better performance compared to the other state-of-the-art change detection methods.
In this paper, we consider Discretized Neural Networks (DNNs) consisting of low-precision weights and activations, which suffer from either infinite or zero gradients caused by the non-differentiable discrete function in the training process. In this case, most training-based DNNs use the standard Straight-Through Estimator (STE) to approximate the gradient w.r.t. discrete value. However, the standard STE will cause the gradient mismatch problem, i.e., the approximated gradient direction may deviate from the steepest descent direction. In other words, the gradient mismatch implies the approximated gradient with perturbations. To address this problem, we introduce the duality theory to regard the perturbation of the approximated gradient as the perturbation of the metric in Linearly Nearly Euclidean (LNE) manifolds. Simultaneously, under the Ricci-DeTurck flow, we prove the dynamical stability and convergence of the LNE metric with the $L^2$-norm perturbation, which can provide a theoretical solution for the gradient mismatch problem. In practice, we also present the steepest descent gradient flow for DNNs on LNE manifolds from the viewpoints of the information geometry and mirror descent. The experimental results on various datasets demonstrate that our method achieves better and more stable performance for DNNs than other representative training-based methods.