In the field of remote sensing image processing, remote sensing image object detection is a crucial undertaking. However, the existing target detection algorithms have a considerable number of model parameters, which results in a slow detection speed that is not conducive to application deployment and real-time detection. Additionally, due to the background complexity of remote sensing images and the large number of small target objects, the performance of existing algorithmic models applied directly with remote sensing images is unsatisfactory. To address the aforementioned issues, this paper proposes an efficient and lightweight architecture based on YOLOv8: AUD-YOLO. In particular, a convolution module, GEConv, based on improved efficient multi-scale attention (EMA), is proposed to replace the original ordinary convolution at the end of the backbone part. This can enhance the accuracy of target detection in remote sensing images while maintaining a lightweight operation. The utilization of dynamic upsampling operators serves to perform upsampling operations with greater accuracy and to more effectively extract image features pertaining to small target objects. Finally, the issue of sample imbalance is addressed by utilising Wise-iou as the position coordinate loss function of the neural network. In the experiments, we validate the effectiveness and robustness of the method using the public remote sensing datasets NWPU VHR-10 and DIOR. The experimental results demonstrate that the AUD-YOLO model proposed in this paper achieves mAP values of 91.1% and 82.7%, which are 1.6% and 1.5% higher than the mAP indexes of the baseline model, and the number of parameters is reduced by 4.7% compared with the original model. We comprehensively consider the number of model parameters and model detection accuracy to make the model more adept at detecting remote sensing images.

The blockchain has been the main computing scenario for smart contracts, and the decentralized infrastructure of the blockchain is effectively implemented in a de-trusted and executable environment. However, vulnerabilities in smart contracts are particularly vulnerable to exploitation by malicious attackers and have always been a key issue in blockchain security. Existing traditional tools are inefficient in detecting vulnerabilities and have a high rate of false positives when detecting contracts. Some neural network methods have improved the detection efficiency, but they are not competent for fine-grained (code line level) vulnerability detection. We proposes the ContractCheck model for detecting contract vulnerabilities based on neural network methods. ContractCheck extracts fine-grained segments from the abstract syntax tree (AST) and function call graph of smart contract source code. Furthermore, the segments are parsed into token flow retaining semantic information as uint, which are used to generate numerical vector sequences that can be trained using neural network methods. We conduct multiple rounds of experiments using a dataset constructed from 36,885 smart contracts and identified the optimal ContractCheck model structure by employing the Fasttext embedding vector algorithm and constructing a composite model using CNN and BiGRU for training the network. Evaluation on other datasets demonstrates that ContractCheck exhibits significant improvement in contract-level detection performance compared to other methods, with an increase of 23.60% in F1 score over the best existing method. Particularly, it achieves fine-grained detection based on neural network methods. The cases provided indicate that ContractCheck can effectively assist developers in accurately locating the presence of vulnerabilities, thereby enhancing the security of Ethereum smart contracts.