Abstract
The importance of recycling is well known, either for environmental or economic reasons, it is impossible to escape it and the industry demands efficiency. Manual labour and traditional industrial sorting techniques are not capable of keeping up with the objectives demanded by the international community. Solutions based in computer vision techniques have the potential automate part of the waste handling tasks. In this paper, we propose a hierarchical deep learning approach for waste detection and classification in food trays. The proposed two-step approach retains the advantages of recent object detectors (as Faster R-CNN) and allows the classification task to be supported in higher resolution bounding boxes. Additionally, we also collect, annotate and make available to the scientific community a new dataset, named Labeled Waste in the Wild, for research and benchmark purposes. In the experimental comparison with standard deep learning approaches, the proposed hierarchical model shows better detection and classification performance. © 2019 IEEE.

Abstract

Abstract

Abstract
Microscopic examination is currently the gold standard test for diagnosis of several neglected tropical diseases. However, reliable identification of parasitic infections requires in-depth train and access to proper equipment for subsequent microscopic analysis. These requirements are closely related with the increasing interest in the development of computer-aided diagnosis systems, and Mobile Health is starting to play an important role when it comes to health in Africa, allowing for distributed solutions that provide access to complex diagnosis even in rural areas. In this paper, we present a 3D-printed microscope that can easily be attached to a wide range of mobile devices models. To the best of our knowledge, this is the first proposed smartphone-based alternative to conventional microscopy that allows autonomous acquisition of a pre-defined number of images at 1000x magnification with suitable resolution, by using a motorized automated stage fully powered and controlled by a smartphone, without the need of manual focus of the smear slide. Reference smears slides with different parasites were used to test the device. The acquired images showed that was possible to visually detect those agents, which clearly illustrate the potential that this device can have, specially in developing countries with limited access to healthcare services.

Abstract

Abstract
Sign Language Recognition (SLR) has become an appealing topic in modern societies because such technology can ideally be used to bridge the gap between deaf and hearing people. Although important steps have been made towards the development of real-world SLR systems, signer-independent SLR is still one of the bottleneck problems of this research field. In this regard, we propose a deep neural network along with an adversarial training objective, specifically designed to address the signer-independent problem. Concretely speaking, the proposed model consists of an encoder, mapping from input images to latent representations, and two classifiers operating on these underlying representations: (i) the signclassifier, for predicting the class/sign labels, and (ii) the signer-classifier, for predicting their signer identities. During the learning stage, the encoder is simultaneously trained to help the sign-classifier as much as possible while trying to fool the signer-classifier. This adversarial training procedure allows learning signer-invariant latent representations that are in fact highly discriminative for sign recognition. Experimental results demonstrate the effectiveness of the proposed model and its capability of dealing with the large inter-signer variations.