Closed Ericjiejie closed 5 years ago
SE blocks can be regarded as a basic component for learning representations with convolutional network architectures. As we know, the module or the idea has been used in a wide range of visual tasks, such as classification, detection, segmentation, depth-estimation and super-resolution though we mainly evaluated it in classification tasks.
Hi,@lishen-shirley Can SE blocks be used for regression tasks? For example, hand posture estimation.