Efficient Medical Image Segmentation Based on Knowledge Distillation.

Recent advances have been made in applying convolutional neural networks to achieve more precise prediction results for medical image segmentation problems. However, the success of existing methods has highly relied on huge computational complexity and massive storage, which is impractical in the real-world scenario. To deal with this problem, we propose an efficient architecture by distilling knowledge from well-trained medical image segmentation networks to train another lightweight network. This architecture empowers the lightweight network to get a significant improvement on segmentation capability while retaining its runtime efficiency. We further devise a novel distillation module tailored for medical image segmentation to transfer semantic region information from teacher to student network. It forces the student network to mimic the extent of difference of representations calculated from different tissue regions. This module avoids the ambiguous boundary problem encountered when dealing with medical imaging but instead encodes the internal information of each semantic region for transferring. Benefited from our module, the lightweight network could receive an improvement of up to 32.6% in our experiment while maintaining its portability in the inference phase. The entire structure has been verified on two widely accepted public CT datasets LiTS17 and KiTS19. We demonstrate that a lightweight network distilled by our method has non-negligible value in the scenario which requires relatively high operating speed and low storage usage.

A color based face detection system using multiple templates.

A color based system using multiple templates was developed and implemented for detecting human faces in color images. The algorithm consists of three image processing steps. The first step is human skin color statistics. Then it separates skin regions from non-skin regions. After that, it locates the frontal human face(s) within the skin regions. In the first step, 250 skin samples from persons of different ethnicities are used to determine the color distribution of human skin in chromatic color space in order to get a chroma chart showing likelihoods of skin colors. This chroma chart is used to generate, from the original color image, a gray scale image whose gray value at a pixel shows its likelihood of representing the skin. The algorithm uses an adaptive thresholding process to achieve the optimal threshold value for dividing the gray scale image into separate skin regions from non skin regions. Finally, multiple face templates matching is used to determine if a given skin region represents a frontal human face or not. Test of the system with more than 400 color images showed that the resulting detection rate was 83%, which is better than most color-based face detection systems. The average speed for face detection is 0.8 second/image (400 x 300 pixels) on a Pentium 3 (800MHz) PC.