At present， most accelerated Magnetic Resonance Imaging （MRI） reconstruction algorithms reconstruct undersampled amplitude images and use real-value convolution for feature extraction， without considering that the MRI data itself is complex， which limits the feature extraction ability of MRI complex data. In order to improve the feature extraction ability of single slice MRI complex data， and thus reconstruct single slice MRI images with clearer details， a Complex Convolution Dual-Domain Cascade Network （ComConDuDoCNet） was proposed. The original undersampled MRI data was used as input， and Residual Feature Aggregation （RFA） blocks were used to alternately extract the dual domain features of the MRI data， ultimately reconstructing the Magnetic Resonance （MR） images with clear texture details. Complex convolution was used as a feature extractor for each RFA block. Different domains were cascaded through Fourier transform or inverse transform， and data consistency layer was added to achieve data fidelity. A large number of experiments were conducted on publicly available knee joint dataset. The comparison results with the Dual-task Dual-domain Network （DDNet） under three different sampling masks with a sampling rate of 20% show that： under the two-dimensional Gaussian sampling mask， the proposed algorithm decreases Normalized Root Mean Square Error （NRMSE） by 13.6%， increases Peak Signal-to-Noise Ratio （PSNR） by 4.3%， and increases Structural SIMilarity （SSIM） by 0.8%； under the Poisson sampling mask， the proposed algorithm decreases NRMSE by 11.0%， increases PSNR by 3.5%， and increases SSIM by 0.1%； under the radial sampling mask， the proposed algorithm decreases NRMSE by 12.3%， increases PSNR by 3.8%， and increases SSIM by 0.2%. The experimental results show that ComConDuDoCNet， combined with complex convolution and dual-domain learning， can reconstruct MR images with clearer details and more realistic visual effects.

A method based on Siamese network and Transformer was proposed to address the low accuracy problem of infrared dim small target tracking. First， a multi-feature extraction cascading moduling was constructed to separately extract the deep features of the infrared dim small target template frame and the search frame， and concatenate them with their corresponding HOG features at the dimension level. Second， a multi-head attention mechanism Transformer was introduced to perform cross-correlation operations between the template feature map and the search feature map， generating a response map. Finally， the target’s center position in the image and the regression bounding box were obtained through the response map upsampling network and bounding box prediction network to complete the tracking of the infrared dim small targets. Test results on a dataset of 13 655 infrared images show that compared with KeepTrack tracking method， the success rate is improved by 5.9 percentage points and the precision is improved by 1.8 percentage points； compared with TransT （Transformer Tracking） method， the success rate is improved by 14.2 percentage points and the precision is improved by 14.6 percentage points. The proposed method is proved to be more accurate in tracking infrared dim small targets in complex backgrounds.

Deep learning based algorithms such as YOLO （You Only Look Once） and Faster Region-Convolutional Neural Network （Faster R-CNN） require a huge amount of training data to ensure the precision of the model， and it is difficult to obtain data and the cost of labeling data is high in many scenarios. And due to the lack of massive training data， the detection range is limited. Aiming at the above problems， a few-shot object Detection algorithm based on Siamese Network was proposed， namely SiamDet， with the purpose of training an object detection model with certain generalization ability by using a few annotated images. Firstly， a Siamese network based on depthwise separable convolution was proposed， and a feature extraction network ResNet-DW was designed to solve the overfitting problem caused by insufficient samples. Secondly， an object detection algorithm SiamDet was proposed based on Siamese network， and based on ResNet-DW， Region Proposal Network （RPN） was introduced to locate the interested objects. Thirdly， binary cross entropy loss was introduced for training， and contrast training strategy was used to increase the distinction among categories. Experimental results show that SiamDet has good object detection ability for few-shot objects， and SiamDet improves AP₅₀ by 4.1% on MS-COCO 20-way 2-shot and 2.6% on PASCAL VOC 5-way 5-shot compared with the suboptimal algorithm DeFRCN （Decoupled Faster R-CNN）.

Magnetic Resonance Imaging （MRI） is widely used in the diagnosis of complex diseases because of its non-invasiveness and good soft tissue contrast. Due to the low speed of MRI， most of the acceleration is currently performed by highly undersampled Magnetic Resonance （MR） signals in k-space. However， the representative algorithms often have the problem of blurred details when reconstructing highly undersampled MR images. Therefore， a highly undersampled MR image reconstruction algorithm based on Residual Graph Convolutional Neural nETwork （RGCNET） was proposed. Firstly， auto-encoding technology and Graph Convolutional neural Network （GCN） were used to build a generator. Secondly， the undersampled image was input into the feature extraction （encoder） network to extract features at the bottom layer. Thirdly， the high-level features of MR images were extracted by the GCN block. Fourthly， the initial reconstructed image was generated through the decoder network. Finally， the final high-resolution reconstructed image was obtained through a dynamic game between the generator and the discriminator. Test results on FastMRI dataset show that at 10%， 20%， 30%， 40% and 50% sampling rates， compared with spatial orthogonal attention mechanism based MRI reconstruction algorithm SOGAN（Spatial Orthogonal attention Generative Adversarial Network）， the proposed algorithm decreases 3.5%， 26.6%， 23.9%， 13.3% and 14.3% on Normalized Root Mean Square Error （NRMSE）， increases 1.2%， 8.7%， 6.9%， 2.9% and 3.2% on Peak Signal-to-Noise Ratio （PSNR） and increases 0.8%， 2.9%， 1.5%， 0.5% and 0.5% on Structural SIMilarity （SSIM） respectively. At the same time， subjective observation also proves that the proposed algorithm can preserve more details and have more realistic visual effects.

Focused on the issue that huge modal difference between cross-modal person re-identification images， pixel alignment and feature alignment are commonly utilized by most of the existing methods to realize image matching. In order to further improve the accuracy of matching two modal images， a multi-input dual-stream network model based on dynamic dual-attention mechanism was designed. Firstly， the neural network was able to learn sufficient feature information in a limited number of samples by adding images of the same person taken by different cameras in each training batch. Secondly， the gray-scale image obtained by homogeneous augmentation was used as an intermediate bridge to retain the structural information of the visible light images and eliminate the color information at the same time. The use of gray-scale images weakened the network’s dependence on color information， thereby strengthening the network model’s ability to mine structural information. Finally， a Weighted Six-Directional triple Ranking （WSDR） loss suitable for images three modalities was proposed， which made full use of cross-modal triple relationship under different angles of view， optimized relative distance between multiple modal features and improved the robustness to modal changes. Experimental results on SYSU-MM01 dataset show that the proposed model increases evaluation indexes Rank-1 and mean Average Precision （mAP） by 4.66 and 3.41 percentage points respectively compared to Dynamic Dual-attentive AGgregation （DDAG） learning model.