Brands Recognition

The brand recognition model recognizes brand masks on the visual source and can be used to calculate brand statistics with DeepLabv3+. Using deeplabv3+, we can automatically label brand images. The recognition model is evaluated on the provided datasets, achieving high accuracy comparable to or better than previous reported results. BlendFeature: Visual Attention Through Multi-Scale Context Encoding *** The proposed framework can not only derive a great representation of visual input but also result in visual outputs that are more interesting and attractive