Active Millimeter Wave (AMMW) imaging technique has been widely applied in the security industries due to its under-controlled privacy concerns and no health hazards. In fact, how to automatically and precisely detect the concealed objects on the human body in the AMMW images is one key issue in industrial security scanner systems. In this paper, we first investigate the deep learning based object detection approaches for the AMMW images, and then we develop a concealed object detection system for the security system in the airport. For our particular application, the concealed objects include several small items. However, most of the current deep learning based methods focus mainly on detecting large objects resulting in the unsatisfactory performance on small objects. To address this issue, we first skip the down-sampling of the last pooling layer and apply the dilated convolution for the following convolutional layers. Then, we propose a context embedding object detection network. Besides, we build a new large-scale dataset to benchmark the concealed object detection tasks in the AMMW images, which includes more than 58k AMMW images. With the proposed algorithm, we finally achieve 90.05% in detection rate and 9.73% in false alarm rate in our dataset.

The active millimeter wave scanner has been widely used for detecting objects concealed underneath a person’s clothing in the field of security inspection and anti-terrorism. However, the active millimeter wave (AMMW) images always suffer from low signal-noise ratio, motion blur, and small size objects, making it challenging to detect concealed objects efficiently and accurately. The scanner usually captures a sequence of images in different views around a human body at once, while the existing algorithms only utilize the single image without considering the relationships among images. In this paper, we design a multi-source aggregation transformer (MATR) with two different attention mechanisms to model spatial correlations within an image and contextual interactions across images. Specifically, a self-attention module is introduced to encode local relationships between the region proposals in each image, while a cross-attention mechanism is built to focus on modeling the cross-correlations between different images. Besides, to handle the problem of small objects in size and suppress the noise in AMMW images, we present a selective context module (SCM). It designs a dynamic selection mechanism to enhance the high-resolution feature with spatial details and make it more distinguishable from the noisy background. Experiments on two AMMW image datasets demonstrate that the proposed methods lead to a remarkable improvement compared to previous state-of-the-art and will benefit the concealed object detection in practice.

Garment transfer aims to transfer the desired garment from a model image with the desired clothing to a target person, which has attracted a great deal of attention due to its wider potential applications. However, considering the model and target persons are often given at different views, body shapes and poses, realistic garment transfer is facing the following challenges that have not been well addressed: 1) deforming the garment; 2) inferring unobserved appearance; 3) preserving fine texture details. To tackle these challenges, we propose a novel SPatial-Aware Texture Transformer (SPATT) model. Different from existing models, SPATT establishes correspondence and infers unobserved clothing appearance by leveraging the spatial prior information of a UV-space. Specifically, the source image is transformed into a partial UV texture map guided by the extracted dense pose. To better infer the unseen appearance utilizing seen region, we first propose a novel coordinate-prior map that defines the spatial relationship between the coordinates in the UV texture map, and design an algorithm to compute it. Based on the proposed coordinate-prior map, we present a novel spatial-aware texture generation network to complete the partial UV texture. In the second stage, we first transform the completed UV texture to fit the target person. To polish the details and improve realism, we introduce a refinement generative network conditioned on the warped image and source input. Compared with existing frameworks as shown experimentally, the proposed framework can generate more realistic images with better-preserved texture details. Furthermore, difficult cases where two persons have large pose and view differences can also be well handled by SPATT.

Fish swarm image segmentation provides an easy to understand and analyze representation for behavior monitoring feature extraction and image information analysis, accurate and effective image segmentation is the basis of fish shoal monitoring. In this paper, algorithm of adaptive fast clustering for fish swarm color image segmentation was proposed by combining the fish swarm hypoxia image features and K-Means++ algorithm. In the RGB color space, the color information of the channel with the maximum average brightness is retained, the others compensation was zero, the generated new image replaced the original. The fish color library was constructed using the gray distribution statistics of the fish swarm targets and background. The pixel probability distribution value in each gray scale range of the normalized gray histogram of the newly generated image is calculated, and combine with the fish group target gray scale reference statistic, the clustering value is determined by two traversing. According to the reserved channel information, the corresponding cluster fish swarm color library is selected for color clustering. The clustering result is processed by threshold transformation to finally realize fish swarm image segmentation. Experiment showed that the average range of structural similarity of our algorithm was [0.93, 1], and the average range of peak signal-to-noise ratio was [44, 50], the running time of the algorithm in this paper is 56% shorter than K-Means ++ algorithm and 71% shorter than the fuzzy clustering algorithm when processing the same images, which could met the requirements of image segmentation quality and accuracy for fish behavior detection.

In the classical k-means algorithm, the value of k must be confirmed in advance. It is difficult to confirm accurately the value of k in reality. This paper proposes an improved genetic k-means algorithm (IGKM) and constructs a fitness function defined as a product of three factors, maximization of which ensures the formation of a small number of compact clusters with large separation between at least two clusters. At last, two artificial and three real-life data sets are considered for experiments that compare IGKM with k-means algorithm, GA-based method and genetic k-means algorithm (GKM) by inter-cluster distance (ITD), inner-cluster distance (IND) and rate of separation exactness. The experiments show that IGKM can automatically reach the optimal value of k with high accuracy