Channel system is intimately linked to the formation of oil and gas reservoirs. In petroliferous basins, channel deposits frequently serve as both storage spaces and fluid conduits. Consequently, accurate identification of channels in 3D seismic data is therefore critical for reservoir prediction. Traditional seismic attribute-based methods can outline channel boundaries, but noise and stratigraphic complexity introduce discontinuities that reduce accuracy and require extensive manual correction. Deep learning-based methods outperform conventional methods in terms of efficiency and precision. However, the similar seismic signatures of channels and continuous karst caves in seismic profiles can still mislead existing models. To address this challenge, we proposed an improved variant of the 3D TransUnet model for 3D seismic data recognition. The model incorporates channel and spatial attention mechanisms into the skip connections of the TransUnet architecture, effectively enhancing its feature representation capability and recognition accuracy. In addition, multi-loss function is introduced to improve the delineation and continuity of channel while increasing the model’s robustness against non-channel interference features. Experiments on synthetic and field seismic data confirm superior boundary delineation, continuity, and noise resistance compared with baseline methods.

The accurate identification and characterization of faults in seismic data is a key step for geological structure interpretation, favorable traps and reservoirs, and recovery of hydrocarbons. Artificial fault marking is not only time-consuming and laborious but also influenced by seismic quality and interpretation experience. The current commonly used U-Net models have satisfactory performance in assisting fault interpretation, but which have limited ability to fine segmentation and recover lost feature information, due to the lack of representative datasets and limitations of network architecture. This paper alleviates this problem by optimizing network architecture, we propose a concise W-Net architecture to identify fault information by designing two expansive paths to obtain rich context information from different scales and enhance the learning ability of the neural network. In the skip connection part, we add double path semantic segmentation features to effectively transfer information, improve the positioning accuracy, optimize memory usage, extract important fault features, and suppress irrelevant information. The synthetic results show that the convergence rate of accuracy and loss of W-Net is faster than that of U-Net in the early stages (epoch<=10). The geologic detailed information of fault identification with W-Net in 3D actual seismic data is more abundant and continuous, which indicates that W-Net can serve as an improved tool to automatically characterize the distribution of geologic discontinuity in seismic images.

Monocular 3D motion capture (mocap) is beneficial to many applications. The use of a single camera, however, often fails to handle occlusions of different body parts and hence it is limited to capture relatively simple movements. We present a light-weight, hybrid mocap technique called HybridCap that augments the camera with only 4 Inertial Measurement Units (IMUs) in a novel learning-and-optimization framework. We first employ a weakly-supervised and hierarchical motion inference module based on cooperative pure residual recurrent blocks that serve as limb, body and root trackers as well as an inverse kinematics solver. Our network effectively narrows the search space of plausible motions via coarse-to-fine pose estimation and manages to tackle challenging movements with high efficiency. We further develop a hybrid optimization scheme that combines inertial feedback and visual cues to improve tracking accuracy. Extensive experiments on various datasets demonstrate HybridCap can robustly handle challenging movements ranging from fitness actions to Latin dance. It also achieves real-time performance up to 60 fps with state-of-the-art accuracy.