Small object detection is one of the key challenges in the current computer vision field due to the low amount of information carried and the information loss caused by feature extraction. You Only Look Once v5 (YOLOv5) adopts the Path Aggregation Network to alleviate the problem of information loss, but it cannot restore the information that has been lost. To this end, an auxiliary information-enhanced YOLO is proposed to improve the sensitivity and detection performance of YOLOv5 to small objects. Firstly, a context enhancement module containing a receptive field size of 21×21 is proposed, which captures the global and local information of the image by fusing multi-scale receptive fields, and introduces an attention branch to enhance the expressive ability of key features and suppress background noise. To further enhance the feature expression ability of small objects, we introduce the high- and low-frequency information decomposed by wavelet transform into PANet to participate in multi-scale feature fusion, so as to solve the problem that the features of small objects gradually disappear after multiple downsampling and pooling operations. Experiments on the challenging dataset Tsinghua–Tencent 100 K show that the mean average precision of the proposed model is 9.5% higher than that of the original YOLOv5 while maintaining the real-time speed, which is better than the mainstream object detection models.
Addressing the challenge that existing deep learning models face in accurately segmenting metal corrosion boundaries and small corrosion areas. In this paper, a SegFormer metal corrosion detection method based on parallel extraction of edge features is proposed. Firstly, to solve the boundary ambiguity problem of metal corrosion images, an edge-feature extraction module (EEM) is introduced to construct a spatial branch of the network to assist the model in extracting shallow details and edge information from the images. Secondly, to mitigate the loss of target feature information during the reconstruction of the decoder, this paper adopts the gradual upsampling decoding layer design. It introduces the feature fusion module (FFM) to achieve hierarchical and progressive feature fusion, thereby enhancing the detection of small corroded areas. Experimental results show that the proposed method outperforms other semantic segmentation models achieving an accuracy of 86.56% on the public metal surface corrosion image dataset and reaching a mean intersection over union (mIoU) of 91.41% on the BSData defect dataset. On the Self-built tubing corrosion pit image dataset, the model utilizes only 3.60 MB of parameters to achieve an accuracy of 96.52%, confirming the effectiveness and performance advantages of the proposed method in practical applications.
To improve the performance of acoustic logging tool in detecting three-dimensional formation, larger and more complicated transducer arrays have been used, which will greatly increase the difficulty of fault diagnosis during tool assembly and maintenance. As a result, traditional passive diagnostic methods become inefficient, and very skilled assemblers and maintainers are required. In this study, fault-diagnosis requirement for the acoustic logging tool at different levels has been analyzed from the perspective of the tool designer. An intelligent fault-diagnosis system consisting of a master-slave hardware architecture and a systemic diagnosis strategy was developed. The hardware system is based on the embedded technology, while the diagnosis strategy is built upon fault-tree analysis and data-driven methods. Diagnostic practice shows that this intelligent system can achieve four levels of fault diagnosis for the acoustic logging tool: System, subsystem, circuit board, and component. This study provided a more rigorous and professional fault diagnosis during tool assembly and maintenance. It is expected that this proposed method would be of great help in achieving cost reduction and improving work efficiency.
The detection depth of current borehole acoustic reflection imaging is only tens of meters without high resolution. This considerably limits its wide application in the identification and fine description of unconventional reservoirs and in the optimization of drilling trajectories. Increasing the directional energy from the transmitter to a geological structure is an excellent way to solve this issue. In this study, a plasma source with a parabolic reflector was introduced during borehole acoustic reflection imaging. First, an experimental system was built for testing the plasma source. Next, the acoustic-electrical characteristics and directional radiation of the source were studied using experiments and a numerical simulation. Finally, the advantages, disadvantages, and feasibility of the plasma-transmitting source were analyzed; some suggestions for further work on the source and its logging application were proposed. The experimental and simulation results show that the use of a plasma source with a parabolic reflector can increase the detection depth of borehole acoustic reflection imaging to hundreds of meters with high resolution. This is crucial in imaging the geological structures near boreholes and enhancing oil–gas exploration and development.
1 January 2016 in Proceedings of Meetings on Acoustics
Phased combined arc array technology was adopted when we designed the acoustic sonde to enhance 3D detection capability and resolution in borehole azimuthal reflection logging tool. After constructing the measurement system, horizontal and vertical radiation characteristics of the transmitter sonde with different phased excitation delays were tested in a 5 m * 5 m * 4 m water tank as well as the consistencies of 80 acquisition channels in 10 stations of the receiver sonde. Also, this tool was placed in open waters to explore the response of reflector including size, placed radial distance and azimuth with optional phased excitation delay. Experimental results show that when the phased excitation delay is 30 us, the main lobe direction of vertical radiation energy of the transmitter could be up-deflected by about 10° with less side lobe energy in water; All channels have satisfactory consistencies and no correction is needed. This tool possesses the ability of recognizing a 2 m * 1 m * 0.05 m reflector at 10 m radial distance and in the expected direction in water, which means about 30 m of the detection capability in real formation. This work is helpful for acquiring performance of the tool and data processing.
1 September 2025 in Review of Scientific Instruments
Currently, notable difficulties exist regarding the real-time uploading of data and fast logging in remote-detection acoustic logging, which can be mitigated via downhole data compression. This study systematically analyzed a wavelet transform–based data compression method and developed hardware platforms based on a digital signal processor (DSP) and field programmable gate array (FPGA). The wavelet transform–based acoustic-logging-data compression algorithm was executed on both the hardware platforms, and the corresponding decompression algorithm was implemented on the host computer. The performance and applicability of the algorithm were evaluated using actual acoustic logging data. Results indicated that the compression ratio and distortion rate of the single-layer wavelet transform–based data compression algorithm exhibited minimal relation with the two hardware platforms. The compression ratio was ∼50%; the reconstructed full waveform effectively preserved the overall morphology of the original signal; and distortions at individual positions exerted negligible impact on the extraction of the sliding longitudinal wave, sliding transverse wave, and reflected wave in the full-waveform data. The wavelet transform–based data compression algorithm occupied less memory in the FPGA platform for processing 2048-word acoustic logging full-waveform data. The execution time was ∼42 μs, which was substantially less than the millisecond-scale runtime required by the DSP platform. This study provides an idea of data compression at the receiver node in a remote-detection acoustic logging tool, which can reduce the workload of the master controller and improve cable transmission and logging efficiencies, serving as a reference for designing next-generation remote-detection acoustic logging tools.