In Currently, research in the field of infrared road object detection is primarily focused on enhancing model performance and robustness to address the challenges posed by complex real-world driving scenarios. In resp...
详细信息
Person Re-Identification falls within the scope of computer vision, acting a technique to ascertain the presence of a specified pedestrian within a video or image library. The related research is of great significance...
详细信息
With the popularity of mobile payment, the problem of financial fraud is becoming increasingly prominent, and the limitations of traditional anti-fraud methods make it more difficult to fight fraud. Many fraud detecti...
详细信息
Most current Visual Question Answering (VQA) methods struggle to achieve effective cross-modal interaction between visual and semantic information, resulting in difficulties in accurately combining visual content with...
详细信息
Low-light image enhancement is highly desirable for outdoor image processing and computer vision applications. Research conducted in recent years has shown that images taken in low-light conditions often pose two main...
详细信息
Underwater target detection is an important part of marine exploration. However, in complex underwater environments due to factors like light absorption and scattering, as well as variations in water quality and clari...
详细信息
As an essential tool for realistic description of the current or future debris environment,the Space Debris Environment engineering Model(SDEEM)has been developed to provide support for risk assessment of *** contrast...
详细信息
As an essential tool for realistic description of the current or future debris environment,the Space Debris Environment engineering Model(SDEEM)has been developed to provide support for risk assessment of *** contrast with SDEEM2015,SDEEM2019,the latest version,extends the orbital range from the Low Earth Orbit(LEO)to Geosynchronous Orbit(GEO)for the years *** this paper,improved modeling algorithms used by SDEEM2019 in propagating simulation,spatial density distribution,and spacecraft flux evaluation are *** debris fluxes of SDEEM2019 are compared with those of three typical models,i.e.,SDEEM2015,Orbital Debris engineering Model 3.1(ORDEM 3.1),and Meteoroid and Space Debris Terrestrial Environment Reference(MASTER-8),in terms of two assessment *** orbital cases,including the Geostationary Transfer Orbit(GTO),Sun-Synchronous Orbit(SSO)and International Space Station(ISS)orbit,are selected for the spacecraft assessment mode,and the LEO region is selected for the spatial density assessment *** analysis indicates that compared with previous algorithms,the variable step-size orbital propagating algorithm based on semi-major axis control is more precise,the spatial density algorithm based on the second zonal harmonic of the non-spherical Earth gravity(J_(2))is more applicable,and the result of the position-centered spacecraft flux algorithm is more *** comparison shows that SDEEM2019 and MASTER-8 have consistent trends due to similar modeling processes,while the differences between SDEEM2019 and ORDEM 3.1 are mainly caused by different modeling approaches for uncatalogued debris.
Partial-label learning(PLL) is a typical problem of weakly supervised learning, where each training instance is annotated with a set of candidate labels. Self-training PLL models achieve state-of-the-art performance b...
详细信息
Partial-label learning(PLL) is a typical problem of weakly supervised learning, where each training instance is annotated with a set of candidate labels. Self-training PLL models achieve state-of-the-art performance but suffer from error accumulation problems caused by mistakenly disambiguated instances. Although co-training can alleviate this issue by training two networks simultaneously and allowing them to interact with each other, most existing co-training methods train two structurally identical networks with the same task, i.e., are symmetric, rendering it insufficient for them to correct each other due to their similar limitations. Therefore, in this paper, we propose an asymmetric dual-task co-training PLL model called AsyCo,which forces its two networks, i.e., a disambiguation network and an auxiliary network, to learn from different views explicitly by optimizing distinct tasks. Specifically, the disambiguation network is trained with a self-training PLL task to learn label confidence, while the auxiliary network is trained in a supervised learning paradigm to learn from the noisy pairwise similarity labels that are constructed according to the learned label confidence. Finally, the error accumulation problem is mitigated via information distillation and confidence refinement. Extensive experiments on both uniform and instance-dependent partially labeled datasets demonstrate the effectiveness of AsyCo.
Objective: In order to improve the friction-increasing and wear-reducing performance of the unfolding wheel surface, the surface microstructure of the unfolding wheel used in the detection of 8 kinds of steel balls wa...
详细信息
In recent years, the utilization of unmanned aerial vehicles (UAVs) for aerial target detection has gained significant attention due to their high-altitude perspective and maneuverability, which offer novel opportunit...
详细信息
暂无评论