Abstract: An energy conversion apparatus and a vehicle are provided. The energy conversion apparatus includes a motor coil of a motor (101), a bridge arm converter (102), a bus capacitor (103) connected to the bridge arm converter (102) in parallel, and a controller (104) connected to the bridge arm converter (102). When the energy conversion apparatus is connected to an external power supply, according to to-be-driven power of the motor and to-be-charged power of an external battery (105), the controller (104) controls the bridge arm converter (102) to cause electrical energy of the external power supply to flow to a drive-charging circuit, and adjusts a current of the drive-charging circuit, to cause the external power supply to drive the motor to output drive power and charge the external battery (105) at the same time.

Abstract: The present application provides a low-temperature process to reduce S—S and/or S—H bonds in organic compounds, including sulfur-cured elastomers, which for example, permits the de-crosslinking of the elastomer and recovery of organic polymers from inorganic constituents.

Abstract: The invention relates to a method for person re-identification based on deep model with multi-loss fusion training strategy. The method uses a deep learning technology to perform preprocessing operations such as flipping, clipping, random erasing and style transfer, and then feature extraction is performed through a backbone network model; joint training of a network is performed by fusing a plurality of loss functions. Compared with other deep learning-based person re-identification algorithms, the present invention greatly improves the performance of person re-identification by adopting a plurality of preprocessing modes, the fusion of three loss functions and effective training strategy.

Abstract: The invention relates to a method for person re-identification based on deep model with multi-loss fusion training strategy. The method uses a deep learning technology to perform preprocessing operations such as flipping, clipping, random erasing and style transfer, and then feature extraction is performed through a backbone network model; joint training of a network is performed by fusing a plurality of loss functions. Compared with other deep learning-based person re-identification algorithms, the present invention greatly improves the performance of person re-identification by adopting a plurality of preprocessing modes, the fusion of three loss functions and effective training strategy.