Smuggling detection system

1. A smuggling detection system, comprising:

a camera configured to capture an input image of a subject purported to be a baby;

a memory storing a deep learning model configured to perform a baby detection task for a smuggling detection application; and

a processor configured to apply the deep learning model to the input image to provide a baby detection result of either a presence or an absence of an actual baby in relation to the subject purported to be the baby, and wherein the baby detection task is configured to evaluate one or more different distractor modalities corresponding to one or more different physical spoofing materials to prevent baby spoofing for the baby detection task.

2. The smuggling detection system of claim 1 , wherein the processor applies the deep learning model to the input image in a multi-task learning process that jointly optimizes an object bounding task and the baby detection task.

3. The smuggling detection system of claim 1 , wherein the deep learning engine model includes a training portion and an architecture portion, and wherein the architecture portion includes a Scale Dependent Pooling detector to enhance baby detection accuracy.

4. The smuggling detection system of claim 1 , wherein the deep learning model includes a training portion and an architecture portion, and wherein the architecture portion is configured to apply a Singular Value Decomposition process that reduces a number of filters in at least some layers of the deep learning model.

5. The smuggling detection system of claim 4 , wherein the Singular Value Decomposition Process is a Truncated Singular Value Decomposition process.

6. The smuggling detection system of claim 4 , wherein all layers, including convolutional layers, of the deep learning model are compressed using the Singular Value Decomposition process.

7. The smuggling detection system of claim 1 , wherein the processor is further configured to generate an alarm indicating the presence or the absence of the actual baby in relation to the subject purported to be the baby, responsive to a result of the baby detection task.

8. The smuggling detection system of claim 1 , wherein the processor is further configured to log a detection of the absence of the actual baby to provide a possible smuggling occurrence count, responsive to a result of the baby detection task.

9. The smuggling baby detection system of claim 1 , wherein the processor is further configured to selectively open a gate to permit access or close the gate to block access, responsive to a result of the baby detection task.

10. The smuggling detection system of claim 1 , wherein the one or more different spoofing materials comprise images of baby dolls.

11. The smuggling detection system of claim 10 , wherein the baby dolls in the images are depicted in various different positions.

12. The smuggling detection system of claim 1 , wherein the processor is further configured to perform data preprocessing on the plurality of input images selected from the group consisting of image contrast enhancements, data augmentation, and cropping.

13. The smuggling detection system of claim 1 , wherein the baby detection task is configured to detect the presence or the absence of the actual baby in a position where legs of the actual baby are off of the ground.

14. A computer-implemented method for smuggling detection, comprising:

capturing, by a camera, an input image of a subject purported to be a baby;

storing, in a memory, a deep learning model configured to perform a baby detection task for a smuggling detection application; and

applying, by a processor, the deep learning model to the input image to provide a baby detection result of either a presence or an absence of an actual baby in relation to the subject purported to be the baby, and wherein the baby detection task is configured to evaluate one or more different distractor modalities corresponding to one or more different physical spoofing materials to prevent baby spoofing for the baby detection task.

15. The smuggling detection system of claim 14 , wherein the processor applies the deep learning model to the input image in a multi-task learning process that jointly optimizes an object bounding task and the baby detection task.

16. The smuggling detection system of claim 14 , wherein the deep learning model includes a training portion and an architecture portion, and wherein the architecture portion includes a Scale Dependent Pooling detector to enhance baby detection accuracy.

17. The smuggling detection system of claim 14 , wherein the deep learning model includes a training portion and an architecture portion, and wherein the architecture portion is configured to apply a Singular Value Decomposition process that reduces a number of filters in at least some layers of the deep learning model.

18. The smuggling detection system of claim 17 , wherein the Singular Value Decomposition Process is a Truncated Singular Value Decomposition process.

19. The smuggling detection system of claim 17 , wherein all layers, including convolutional layers, of the deep learning model are compressed using the Singular Value Decomposition process.

20. A computer program product for smuggling detection, the computer program product comprising a non-transitory computer readable storage medium having program instructions embodied therewith, the program instructions executable by a computer to cause the computer to perform a method comprising:

capturing, by a camera, an input image of a subject purported to be a baby;

storing, in a memory, a deep learning model configured to perform a baby detection task for a smuggling detection application; and

applying, by a processor, the deep learning model to the input image to provide a baby detection result of either a presence or an absence of an actual baby in relation to the subject purported to be the baby, and wherein the baby detection task is configured to evaluate one or more different distractor modalities corresponding to one or more different physical spoofing materials to prevent baby spoofing for the baby detection task.