Object detection is a significant downstream task in computer vision. For the on-board edge computing platforms, a giant model is difficult to achieve the real-time detection requirement. And, a lightweight model built from a large number of the depth-wise separable convolution layers cannot achieve the sufficient accuracy. We introduce a new lightweight convolution technique, GSConv, to lighten the model but maintain the accuracy. The GSConv accomplishes an excellent trade-off between the model's accuracy and speed. And, we provide a design paradigm, slim-neck, to achieve a higher computational cost-effectiveness of the detectors. The effectiveness of our approach was robustly demonstrated in over twenty sets comparative experiments. In particular, the detectors of ameliorated by our approach obtains state-of-the-art results (e.g. 70.9% mAP0.5 for the SODA10M at a speed of ~ 100FPS on a Tesla T4 GPU) compared with the originals. Code is available at https://github.com/alanli1997/slim-neck-by-gsconv

In this study, the effect of Ce additions on microstructure evolution of Mg–7Gd–3.5Y–0.3Zn (wt %) alloys during the casting, homogenization, aging and extrusion processing are investigated, and novel mechanical properties are also obtained. The results show that Ce addition promotes the formation of long period stacking ordered (LPSO) phases in the as-cast Mg–Gd–Y–Zn–Ce alloys. A high content of Ce addition would reduce the maximum solubility of Gd and Y in the Mg matrix, which leads to the higher density of Mg12Ce phases in the as-homogenized alloys. The major second phases observed in the as-extruded alloys are micron-sized bulk LPSO phases, nano-sized stripe LPSO phases, and broken Mg12Ce and Mg5RE phases. Recrystallized grain size of the as-extruded 0.2Ce, 0.5Ce and 1.0Ce alloys can be refined to ~4.3 μm, ~1.0 μm and ~8.4 μm, respectively, which is caused by the synthesized effect of both micron phases and nano phases. The strength and ductility of as-extruded samples firstly increase and then decrease with increasing Ce content. As-extruded 0.5Ce alloy exhibits optimal mechanical properties, with ultimate strength of 365 MPa and ductility of ~15% simultaneously.

Pedestrian detection is one of the most widely applied tasks in industrial computer vision. It encapsulates three core challenges of object detection: detecting small objects, handling heavy occlusion, and balancing speed and accuracy for deployment on mobile devices. In targeting scenarios relating to the Internet of Things (IoT), we propose a dedicated lightweight pedestrian detector that is robust to occlusions. First, we redesign the decoupled prediction head with a hierarchical structure, separating classification confidence estimation from bounding box regression. We then decode the offsets from the regression branch, extract features from high-confidence predictions, and fuse these with classification feature maps to enhance the local reliability of semantic features. Furthermore, we introduce a label-dynamic matching strategy that increases the number of high-quality positive samples, particularly improving matching for small and occluded objects. Finally, an optimized knowledge distillation framework significantly boosts the prediction accuracy of the compact model, facilitating deployment on edge devices. Experimental results on the CrowdHuman test set show that our proposed approach achieves comparable accuracy to the baseline (53.8 %) with an inference latency of only 7.1 ms – 281.7 % faster than the baseline.