Recent advances of Reinforcement Learning (RL) have highlighted its potential in complex reasoning tasks, yet effective training often relies on external supervision, which limits the broader applicability. In this work, we propose a novel self-rewarding reinforcement learning framework to enhance Large Language Model (LLM) reasoning by leveraging the consistency of intermediate reasoning states across different reasoning trajectories. Our key insight is that correct responses often exhibit consistent trajectory patterns in terms of model likelihood: their intermediate reasoning states tend to converge toward their own final answers (*high consistency*) with minimal deviation toward other candidates (*low volatility*). Inspired by this observation, we introduce CoVo, an intrinsic reward mechanism that integrates ***Co***nsistency and ***Vo***latility via a robust vector-space aggregation strategy, complemented by a curiosity bonus to promote diverse exploration. CoVo enables LLMs to perform RL in a self-rewarding manner, offering a scalable pathway for learning to reason without external supervision. Extensive experiments on diverse reasoning benchmarks show that CoVo achieves performance comparable to or even surpassing supervised RL. Our code is available at https://github.com/sastpg/CoVo.

Weakly-supervised video anomaly detection (WS-VAD) aims to identify fine-grained anomalies from sparse video-level labels, which has gained increasing attention in recent years due to its various applications such as disaster warning and public security. Recent studies typically formulate WS-VAD as a multi-instance learning (MIL) problem. However, they neglect the instance creation process and simply apply a uniform temporal pooling (UTP) operation to obtain the training instances, leading to severe anomaly contamination and dilution. In this paper, we emphasize the importance of the instance modeling procedure and propose two simple yet effective modules, i.e., the dynamic segment merging (DSM) module and the retrieval-augmented anomaly restoration (RA2R) module, to tackle the problem from segment-level and feature-level, respectively. We equip various state-of-the-art WS-VAD models with the proposed methods and conduct thorough experiments on the challenging datasets, e.g., UCF-Crime, and XD-Violence. Results demonstrate the proposed method brings consistent performance improvement and establishes new state-of-the-art.

Landing on a vertically oscillating platform poses a significant challenge for multi-rotor unmanned aerial vehicle (UAVs) due to the time-varying ground effect (GE). In this work, we formulated a data-driven GE dynamic model that accurately describes the complex interactions between UAVs and both stationary and oscillating platforms. Integrating this model with a feedforward controller effectively compensates for GE, resulting in improved landing performance. The proposed GE model elucidates the relationship between GE and factors such as UAVs’ velocity, throttle magnitude, and the motion of the landing platform. It highlights that the GE experienced during the landing process of UAVs is not only contingent on the current state but also related to past states. The resulting GE model is parsimonious and suitable for onboard computers with limited computational power, and its accuracy has been confirmed through a series of flight experiments. To demonstrate the effectiveness of the developed UAVs landing scheme, we compared our approach with robust control and internal model control methods. Experimental results indicate that the proposed landing strategy achieves faster and smoother landings, with at least a 22% improvement in smoothness and a 25% reduction in landing time.

Structural re-parameterization has drawn increasing attention in various computer vision tasks. It aims at improving the performance of deep models without introducing any inference-time cost. Though efficient during inference, such models rely heavily on the complicated training-time blocks to achieve high accuracy, leading to large extra training cost. In this paper, we present online convolutional re-parameterization (OREPA), a two-stage pipeline, aiming to reduce the huge training overhead by squeezing the complex training-time block into a single convolution. To achieve this goal, we introduce a linear scaling layer for better optimizing the online blocks. Assisted with the reduced training cost, we also explore some more effective re-param components. Compared with the state-of-the-art re-param models, OREPA is able to save the training-time memory cost by about 70% and accelerate the training speed by around 2×. Meanwhile, equipped with OREPA, the models out-perform previous methods on ImageNet by up to +0.6%. We also conduct experiments on object detection and semantic segmentation and show consistent improvements on the downstream tasks. Codes are available at https://github.com/JUGGHM/OREPA_CVPR2022.

Recently, unsupervised person re-identification (Re-ID) has received increasing research attention due to its potential for label-free applications. A promising way to address unsupervised Re-ID is clustering-based, which generates pseudo labels by clustering and uses the pseudo labels to train a Re-ID model iteratively. However, most clustering-based methods take each cluster as a pseudo identity class, neglecting the intra-cluster variance mainly caused by the change of cameras. To address this issue, we propose to split each single cluster into multiple proxies according to camera views. The camera-aware proxies explicitly capture local structures within clusters, by which the intra-ID variance and inter-ID similarity can be better tackled. Assisted with the camera-aware proxies, we design two proxy-level contrastive learning losses that are, respectively, based on offline and online association results. The offline association directly associates proxies according to the clustering and splitting results, while the online strategy dynamically associates proxies in terms of up-to-date features to reduce the noise caused by the delayed update of pseudo labels. The combination of two losses enables us to train a desirable Re-ID model. Extensive experiments on three person Re-ID datasets and one vehicle Re-ID dataset show that our proposed approach demonstrates competitive performance with state-of-the-art methods. Code will be available at: https://github.com/Terminator8758/O2CAP.

This paper tackles the purely unsupervised person re-identification (Re-ID) problem that requires no annotations. Some previous methods adopt clustering techniques to generate pseudo labels and use the produced labels to train Re-ID models progressively. These methods are relatively simple but effective. However, most clustering-based methods take each cluster as a pseudo identity class, neglecting the large intra-ID variance caused mainly by the change of camera views. To address this issue, we propose to split each single cluster into multiple proxies and each proxy represents the instances coming from the same camera. These camera-aware proxies enable us to deal with large intra-ID variance and generate more reliable pseudo labels for learning. Based on the camera-aware proxies, we design both intra and inter-camera contrastive learning components for our Re-ID model to effectively learn the ID discrimination ability within and across cameras. Meanwhile, a proxy-balanced sampling strategy is also designed, which facilitates our learning further. Extensive experiments on three large-scale Re-ID datasets show that our proposed approach outperforms most unsupervised methods by a significant margin. Especially, on the challenging MSMT17 dataset, we gain 14.3 percent Rank-1 and 10.2 percent mAP improvements when compared to the second place. Code is available at: https://github.com/Terminator8758/CAP-master.

Intra-camera supervision (ICS) for person reidentification (Re-ID) assumes that identity labels are independently annotated within each camera view and no inter-camera identity association is labeled. It is a new setting proposed recently to reduce the burden of annotation while expect to maintain desirable Re-ID performance. However, the lack of inter-camera labels makes the ICS Re-ID problem much more challenging than the fully supervised counterpart. By investigating the characteristics of ICS, this paper proposes jointly learned camera-specific non-parametric classifiers, together with a hybrid mining quintuplet loss, to perform intra-camera learning. Then, an inter-camera learning module consisting of a graph-based ID association step and a Re-ID model updating step is conducted. Extensive experiments on three large-scale Re-ID datasets show that our approach outperforms all existing ICS works by a great margin. Our approach performs even comparable to state-of-the-art fully supervised methods in two of the datasets.

Weakly supervised object detection (WSOD), which is the problem of learning detectors using only image-level labels, has been attracting more and more interest. However, this problem is quite challenging due to the lack of location supervision. To address this issue, this paper integrates saliency into a deep architecture, in which the location information is explored both explicitly and implicitly. Specifically, we select highly confident object proposals under the guidance of class-specific saliency maps. The location information, together with semantic and saliency information, of the select proposals are then used to explicitly supervise the network by imposing two additional losses. Meanwhile, a saliency prediction sub-network is built in the architecture. The prediction results are used to implicitly guide the localization procedure. The entire network is trained end-to-end. Experiments on PASCAL VOC demonstrate that our approach outperforms all state-of-the-arts.

In this paper, we propose a novel method to perform weakly-supervised image parsing based on the dictionary learning framework. To deal with the challenges caused by the label ambiguities, we design a saliency guided weight assignment scheme to boost the discriminative dictionary learning. More specifically, with a collection of tagged images, the proposed method first conducts saliency detection and automatically infers the confidence for each semantic class to be foreground or background. These clues are then incorporated to learn the dictionaries, the weights, as well as the sparse representation coefficients in the meanwhile. Once obtained the coefficients of a superpixel, we use a sparse representation classifier to determine its semantic label. The approach is validated on the MSRC21, PASCAL VOC07, and VOC12 datasets. Experimental results demonstrate the encouraging performance of our approach in comparison with some state-of-the-arts.