Deep Ensembles Based on Stochastic Activations for Semantic Segmentation

Semantic segmentation is a very popular topic in modern computer vision, and it has applications in many fields. Researchers have proposed a variety of architectures for semantic image segmentation. The most common ones exploit an encoder–decoder structure that aims to capture the semantics of the image and its low-level features. The encoder uses convolutional layers, in general with a stride larger than one, to extract the features, while the decoder recreates the image by upsampling and using skip connections with the first layers. The objective of this study is to propose a method for creating an ensemble of CNNs by enhancing diversity among networks with different activation functions. In this work, we use DeepLabV3+ as an architecture to test the effectiveness of creating an ensemble of networks by randomly changing the activation functions inside the network multiple times. We also use different backbone networks in our DeepLabV3+ to validate our findings. A comprehensive evaluation of the proposed approach is conducted across two different image segmentation problems: the first is from the medical field, i.e., polyp segmentation for early detection of colorectal cancer, and the second is skin detection for several different applications, including face detection, hand gesture recognition, and many others. As to the first problem, we manage to reach a Dice coefficient of 0.888, and a mean intersection over union (mIoU) of 0.825, in the competitive Kvasir-SEG dataset. The high performance of the proposed ensemble is confirmed in skin detection, where the proposed approach is ranked first concerning other state-of-the-art approaches (including HarDNet) in a large set of testing datasets.

PKT: fast color-based spatial model for human skin detection

We present a new detection method for color-based object detection, which can improve the performance of learning procedures in terms of speed, accuracy, and efficiency, using spatial inference, and algorithm. We applied the model to human skin detection from an image; however, the method can also work for other machine learning tasks involving image pixels. We propose (1) an improved RGB/HSL human skin color threshold to tackle darker human skin color detection problem. (2), we also present a new rule-based fast algorithm (packed k-dimensional tree --- PKT) that depends on an improved spatial structure for human skin/face detection from colored 2D images. We also implemented a novel packed quad-tree (PQT) to speed up the quad-tree performance in terms of indexing. We compared the proposed system to traditional pixel-by-pixel (PBP)/pixel-wise (PW) operation, and quadtree based procedures. The results show that our proposed spatial structure performs better (with a very low false hit rate, very high precision, and accuracy rate) than most state-of-the-art models.

Deep ensembles based on Stochastic Activations for Semantic Segmentation

Semantic segmentation is a very popular topic in modern computer vision and it has applications to many fields. Researchers proposed a variety of architectures over time, but the most common ones exploit an encoder-decoder structure that aims to capture the semantics of the image and it low level features. The encoder uses convolutional layers, in general with a stride larger than one, to extract the features, while the decoder recreates the image by upsampling an using skip connections with the first layers. In this work, we use DeepLab as architecture to test the effectiveness of creating an ensemble of networks by randomly changing the activation functions inside the network multiple times. We also use different backbone networks in our DeepLab to validate our findings. We manage to reach a dice coefficient of 0.888, and a mean Intersection over Union (mIoU) of 0.825, in the competitive Kvasir-SEG dataset. Results in skin detection also confirm the performance of the proposed ensemble, which is ranked first with respect to other state-of-the-art approaches (including HardNet) in a large set of testing datasets. The developed code will be available at https://github.com/LorisNanni.