Abstract: A method for detecting and tracking target object in a captured video using convolutional neural network (CNN) is provided. The method includes: inputting image data into a detecting model to generate detection results, wherein the detecting model is constructed by the CNN; inputting the image data into tracking models to generate tracking results; performing detection score enhancement operation according to the detection results and the tracking results to obtain enhanced detection results; matching the enhanced detection results and the tracking results by a matching operation; processing matched results and unmatched target detection results and unmatched target tracking results; and selectively updating the tracking models using tracking reliability estimation according to the matched results.

Abstract: In a convolutional neural network (CNN) using an encoder-decoder structure for image segmentation, a multi-scale context aggregation module receives an encoded final-stage feature map from the encoder, and sequentially aggregates multi-scale contexts of this feature map from a global scale to a local scale to strengthen semantic relationships of contexts of different scales to improve segmentation accuracy. The multi-scale contexts are obtained by computing atrous convolution on the feature map for different dilation rates. To reduce computation, a channel-wise feature selection (CFS) module is used in the decoder to merge two input feature maps. Each feature map is processed by a global pooling layer followed by a fully connected layer or a 1×1 convolutional layer to select channels of high activation. By subsequent channel-wise multiplication and elementwise summation, only channels with high activation in both feature maps are preserved and enhanced in the merged feature map.

Abstract: In a convolutional neural network (CNN) using an encoder-decoder structure for image segmentation, a multi-scale context aggregation module receives an encoded final-stage feature map from the encoder, and sequentially aggregates multi-scale contexts of this feature map from a global scale to a local scale to strengthen semantic relationships of contexts of different scales to improve segmentation accuracy. The multi-scale contexts are obtained by computing atrous convolution on the feature map for different dilation rates. To reduce computation required by the CNN, a channel-wise feature selection (CFS) module is used in the decoder to merge two input feature maps. Each feature map is processed by a global pooling layer followed by a fully connected layer or a 1×1 convolutional layer to select channels of high activation. By subsequent channel-wise multiplication and elementwise summation, only channels with high activation in both feature maps are preserved and enhanced in the merged feature map.