Security in Transformer Visual Trackers: A Case Study on the Adversarial Robustness of Two Models.

Visual object tracking is an important technology in camera-based sensor networks, which has a wide range of practicability in auto-drive systems. A transformer is a deep learning model that adopts the mechanism of self-attention, and it differentially weights the significance of each part of the input data. It has been widely applied in the field of visual tracking. Unfortunately, the security of the transformer model is unclear. It causes such transformer-based applications to be exposed to security threats. In this work, the security of the transformer model was investigated with an important component of autonomous driving, i.e., visual tracking. Such deep-learning-based visual tracking is vulnerable to adversarial attacks, and thus, adversarial attacks were implemented as the security threats to conduct the investigation. First, adversarial examples were generated on top of video sequences to degrade the tracking performance, and the frame-by-frame temporal motion was taken into consideration when generating perturbations over the depicted tracking results. Then, the influence of perturbations on performance was sequentially investigated and analyzed. Finally, numerous experiments on OTB100, VOT2018, and GOT-10k data sets demonstrated that the executed adversarial examples were effective on the performance drops of the transformer-based visual tracking. White-box attacks showed the highest effectiveness, where the attack success rates exceeded 90% against transformer-based trackers.

Out-of-Plane Electronics on Flexible Substrates Using Inorganic Nanowires Grown on High-Aspect-Ratio Printed Gold Micropillars.

Out-of-plane or 3D electronics on flexible substrates are an interesting direction that can enable novel solutions such as efficient bioelectricity generation and artificial retina. However, the development of devices with such architectures is limited by the lack of suitable fabrication techniques. Additive manufacturing (AM) can but often fail to provide high-resolution, sub-micrometer 3D architectures. Herein, the optimization of a drop-on-demand (DoD), high-resolution electrohydrodynamic (EHD)-based jet printing method for generating 3D gold (Au) micropillars is reported. Libraries of Au micropillar electrode arrays (MEAs) reaching a maximum height of 196 µm and a maximum aspect ratio of 52 are printed. Further, by combining AM with the hydrothermal growth method, a seedless synthesis of zinc oxide (ZnO) nanowires (NWs) on the printed Au MEAs is demonstrated. The developed hybrid approach leads to hierarchical light-sensitive NW-connected networks exhibiting favorable ultraviolet (UV) sensing as demonstrated via fabricating flexible photodetectors (PDs). The 3D PDs exhibit an excellent omnidirectional light-absorption ability and thus, maintain high photocurrents over wide light incidence angles (±90°). Lastly, the PDs are tested under both concave and convex bending at 40 mm, showing excellent mechanical flexibility.

Printed n- and p-Channel Transistors using Silicon Nanoribbons Enduring Electrical, Thermal, and Mechanical Stress.

Printing technologies are changing the face of electronics with features such as resource-efficiency, low-cost, and novel form factors. While significant advances have been made in terms of organic electronics, the high-performance and stable transistors by printing, and their large-scale integration leading to fast integrated circuits remains a major challenge. This is because of the difficulties to print high-mobility semiconducting materials and the lack of high-resolution printing techniques. Herein, we present silicon based printed n- and p-channel transistors to demonstrate the possibility of developing high-performance complementary metal-oxide-semiconductor (CMOS) computing architecture. The direct roll transfer printing is used here for deterministic assembly of high-mobility single crystal silicon nanoribbons arrays on a flexible polyimide substrate. This is followed by high-resolution electrohydrodynamic printing to define source/drain/gate electrodes and to encapsulate, thus leading to printed devices. The printed transistors show effective peak mobilities of 15 cm2/(V s) (n-channel) and 5 cm2/(V s) (p-channel) at low 1 V drain bias. Furthermore, the effect of electrical, mechanical, and thermal stress on the performance and stability of the encapsulated transistors is investigated. The transistors showed stable transfer characteristics even after: (i) continuous 4000 transfer cycles, (ii) excruciating 10000 bending cycles at different bending radii (40, 25, and 15 mm), and (iii) between 15 and 60 °C temperatures.

Ionothermal Synthesis of Fully Conjugated Covalent Organic Frameworks for High-Capacity and Ultrastable Potassium-Ion Batteries.

Fully aromatic conjugated covalent organic frameworks (FAC-COFs) with excellent physicochemical stability have been emerging as active semiconductors for diverse potential applications. Developing efficient synthesis methods for fabricating FAC-COFs will significantly facilitate the exploration over their material and photonic/electronic functionalities. Herein, a facile solvent-free strategy is developed for the synthesis of 2D phthalocyanine-based FAC-COFs (FAC-Pc-COFs). Cyclopolymerization of benzo[1,2-b:4,5-b']bis[1,4]benzodioxin-2,3,9,10-tetracarbonitrile (BBTC) and quinoxalino[2',3':9,10]phenanthro[4,5-abc]phenazine-6,7,15,16-tetracarbonitrile (QPPTC) in ZnCl2  leads to the fast formation and isolation of BB-FAC-Pc-COF and QPP-FAC-Pc-COF, respectively. Powder X-ray diffraction and electron microscopy analysis reveal their crystalline nature with sql topology and AA stacking configuration. Thermogravimetric analysis and immersion experiment indicate their excellent stability. The conductivity test demonstrates their high conductivity of 0.93-1.94 × 10-4  S cm-1  owing to the fully π-conjugated electronic structural nature. In particular, the as-prepared FAC-Pc-COFs show high-performance K+ storage in potassium-ion batteries due to their excellent conductivity, highly ordered and robust structure, and N/O-rich framework nature. Impressively, QPP-FAC-Pc-COF shows a large reversible capacity of 424 mA h g-1  after 100 cycles at 50 mA g-1  and a capacity retention of nearly 100% at 2000 mA g-1  for over 10 000 cycles.