Abstract: Disclosed is a method for training shallow convolutional neural networks for infrared target detection using a two-phase learning strategy that can converge to satisfactory detection performance, even with scale-invariance capability. In the first step, the aim is to ensure that only filters in the convolutional layer produce semantic features that serve the problem of target detection. L2-norm (Euclidian norm) is used as loss function for the stable training of semantic filters obtained from the convolutional layers. In the next step, only the decision layers are trained by transferring the weight values in the convolutional layers completely and freezing the learning rate. In this step, unlike the first, the L1-norm (mean-absolute-deviation) loss function is used.

Abstract: A noise elimination method for detection applications to increase performance, robustness and efficiency is provided. The noise elimination method including the following steps of; initialization; candidate generation; obtain sensor data; obtain a transformation matrix; warping candidate signals into a Cumulative Signal Frame (CSF); get a next candidate; process a last candidate; query a set handler; add the candidate signals into the set handler; warp the candidate signals onto the CSF; add the candidate signals to the CSF at a warped position; process the CSF; interpret the CSF and generate target candidates; detect decisions and terminate.

Abstract: Disclosed is a method for training shallow convolutional neural networks for infrared target detection using a two-phase learning strategy that can converge to satisfactory detection performance, even with scale-invariance capability. In the first step, the aim is to ensure that only filters in the convolutional layer produce semantic features that serve the problem of target detection. L2-norm (Euclidian norm) is used as loss function for the stable training of semantic filters obtained from the convolutional layers. In the next step, only the decision layers are trained by transferring the weight values in the convolutional layers completely and freezing the learning rate. In this step, unlike the first, the L1-norm (mean-absolute-deviation) loss function is used.

Abstract: A noise elimination method for detection applications to increase performance, robustness and efficiency is provided. The noise elimination method including the following steps of; initialization; candidate generation; obtain sensor data; obtain a transformation matrix; warping candidate signals into a Cumulative Signal Frame (CSF); get a next candidate; process a last candidate; query a set handler; add the candidate signals into the set handler; warp the candidate signals onto the CSF; add the candidate signals to the CSF at a warped position; process the CSF; interpret the CSF and generate target candidates; detect decisions and terminate.