Search

The shallow feature map of the single-shot detector (SSD) is not always conducive to enhancing the recognition precision for a small object because of the lack of contextual information. In this research, a single-shot detection algorithm based on cyclic attention (CA-SSD) is proposed to construct a fast and accurate detector that efficiently obtains full-image contextual information. Our network is constructed by integrating ResNet-34 and proposed novel cyclic attention blocks. This type of building block aggregates different transformations, one of which includes an attention module that uses a long but narrow pooling kernel to acquire horizontal and vertical contextual information for each pixel of all pixels. Each pixel eventually captures the full-image dependencies by following an even further cyclic operation. Our design considers the variability of the gradient, which not only improves the reliability of the cyclic attention block but also cuts the number of parameters for computation. Additionally, by exploring the effects of the stem block and its stride on the performance of ResNet-based SSD algorithms, our network retains more detailed information. For an input size of 300 $\times $ 300, CA-SSD attained 82.5% mAP on PASCAL VOC 2007 test, 78.4% mAP on PASCAL VOC 2012 test, and 32.7% mAP on MS COCO. Experimental results achieved with CA-SSD surpass the best results achieved with the traditional SSD and other advanced object detection algorithms while real-time speed is maintained.

The stable exchange of energy and the effective conduction of heat are particularly critical to ensuring the stable operation of thermoelectric components in entire thermoelectric conversion processes. Poor heat exchange and conduction will cause thermal energy in a thermoelectric system to accumulate, leading to the consequence of burning the energy conversion modules. To solve these problems, a thermoelectric energy conversion system equipped with a water-cooling thermal energy adjustment structure settled on the surface of the heat source was proposed. The working temperature of the heat source, the output voltage, current, and power of the thermoelectric system with and without the structure were obtained and compared. Through comparative analysis, benefiting from the above structure, the temperature fluctuation of the heat sink surface of the heat source was found to drop from 10.36 K to 1.18 K, and the maximum output of the system increased from 255 to 290 mV through the process of input voltage from 1 to 6 V. Furthermore, in the process of gradually increasing the load from 0 to 180 ohms, the system achieves an increased output of 53.8 mV, 1.81 mA and 52.1 mV, 1.83 mA when the input voltage was 4 and 5 V, respectively. In conclusion, the application and design of the above structure obviously promotes the stability and output capacity during thermoelectric energy conversion. The combination of the water-cooling thermal energy adjustment structure with pulse-width modulation (PWM) or maximum power point tracking (MPPT) theory is worth further study.