Augusto Salazar

Towards automatic wild animal monitoring: Identification of animal species in camera-trap images using very deep convolutional neural networks

Non intrusive monitoring of animals in the wild is possible using camera trapping framework, which uses cameras triggered by sensors to take a burst of images of animals in their habitat. However camera trapping framework produces a high volume of data (in the order on thousands or millions of images), which must be analyzed by a human expert. In this work, a method for animal species identification in the wild using very deep convolutional neural networks is presented. Multiple versions of the Snapshot Serengeti dataset were used in order to probe the ability of the method to cope with different challenges that camera-trap images demand. The method reached 88.9% of accuracy in Top-1 and 98.1% in Top-5 in the evaluation set using a residual network topology. Also, the results show that the proposed method outperforms previous approximations and proves that recognition in camera-trap images can be automated.

Camera-trap images segmentation using multi-layer robust principal component analysis

The segmentation of animals from camera-trap images is a difficult task. To illustrate, there are various challenges due to environmental conditions and hardware limitation in these images. We proposed a multi-layer robust principal component analysis (multi-layer RPCA) approach for background subtraction. Our method computes sparse and low-rank images from a weighted sum of descriptors, using color and texture features as case of study for camera-trap images segmentation. The segmentation algorithm is composed of histogram equalization or Gaussian filtering as pre-processing, and morphological filters with active contour as post-processing. The parameters of our multi-layer RPCA were optimized with an exhaustive search. The database consists of camera-trap images from the Colombian forest taken by the Instituto de Investigación de Recursos Biológicos Alexander von Humboldt. We analyzed the performance of our method in inherent and therefore challenging situations of camera-trap images. Furthermore, we compared our method with some state-of-the-art algorithms of background subtraction, where our multi-layer RPCA outperformed these other methods. Our multi-layer RPCA reached 76.17 and 69.97% of average fine-grained F-measure for color and infrared sequences, respectively. To our best knowledge, this paper is the first work proposing multi-layer RPCA and using it for camera-trap images segmentation.

Animal Identification in Low Quality Camera-Trap Images Using Very Deep Convolutional Neural Networks and Confidence Thresholds

Monitoring animals in the wild without disturbing them is possible using camera trapping framework. Automatic triggered cameras, which take a burst of images of animals in their habitat, produce great volumes of data, but often result in low image quality. This high volume data must be classified by a human expert. In this work a two step classification is proposed to get closer to an automatic and trustfully camera-trap classification system in low quality images. Very deep convolutional neural networks were used to distinguish images, firstly between birds and mammals, secondly between mammals sets. The method reached \(97.5\%\) and \(90.35\%\) in each task. An alleviation mode using a confidence threshold of automatic classification is proposed, allowing the system to reach \(100\%\) of performance traded with human work.

HMM Based Evaluation of Physical Therapy Movements Using Kinect Tracking

Recognition of human activities in videos has experienced considerable changes with the introduction of cost-effective technology that allows for the tracking of individual body parts. This has led to the development of numerous tele-health applications that aim to help patients in their recovery process. Most of these systems are based on techniques to measure the degree of similarity of time series, together with thresholds to evaluate whether the movement satisfies the specification. This means that sequences similar enough to a template, but containing deviations from the correct form, may be considered correct, and thus the quality of movement incorrectly assessed. In this paper we propose the use of Hidden Markov Models as novelty detectors to evaluate the quality of movement in human beings. The results show the potential of this approach in detecting the sequences that deviate from normality for a wide range of activities common in physical therapy and rehabilitation.

Automatic identification of Scenedesmus polymorphic microalgae from microscopic images

Microalgae counting is used to measure biomass quantity. Usually, it is performed in a manual way using a Neubauer chamber and expert criterion, with the risk of a high error rate. Scenedesmus algae can build coenobia consisting of 1, 2, 4 and 8 cells. The amount of algae of each coenobium helps to determine the amount of lipids, proteins, and other substances in a given sample of a algae crop. The knowledge of the quantity of those elements improves the quality of bioprocess applications. This paper addresses the methodology for automatic identification of Scenedesmus microalgae (used in the methane production and food industry) and applies it to images captured by a digital microscope. The use of contrast adaptive histogram equalization for pre-processing, and active contours for segmentation are presented. The calculation of statistical features (histogram of oriented gradients, Hu and Zernike moments) with texture features (Haralick and local binary patterns descriptors) are proposed for algae characterization. Classification of coenobia achieves accuracies of 98.63% and 97.32% with support vector machine and artificial neural network, respectively. According to the results, it is possible to consider the proposed methodology as an alternative to the traditional technique for algae counting. In addition, the database used for the developing of the proposed methodology is publicly available.

Automatic Detection of Deviations in Human Movements Using HMM: Discrete vs Continuous

Automatic detection of correct performance of movements in humans is the core of coaching and rehabilitation applications. Human movement can be studied in terms of sequential data by using different sensor technologies. This representation makes it possible to use models that use sequential data to determine if executions of a certain activity are close enough to the specification or if they must be considered to be erroneous. One of the most widely used approaches for characterization of sequential data are Hidden Markov Models (HMM). They have the advantage of being able to model processes based on data from noisy sources. In this work we explore the use of both discrete and continuous HMMs to label movement sequences as either according to a specification or deviated from it. The results show that the majority of sequences are correctly labeled by the technique, with an advantage for continuous HMM.

A Markov Random Field and Active Contour Image Segmentation Model for Animal Spots Patterns

Non-intrusive biometrics of animals using images allows to analyze phenotypic populations and individuals with patterns like stripes and spots without affecting the studied subjects. However, non-intrusive biometrics demand a well trained subject or the development of computer vision algorithms that ease the identification task. In this work, an analysis of classic segmentation approaches that require a supervised tuning of their parameters such as threshold, adaptive threshold, histogram equalization, and saturation correction is presented. In contrast, a general unsupervised algorithm using Markov Random Fields (MRF) for segmentation of spots patterns is proposed. Active contours are used to boost results using MRF output as seeds. As study subject the Diploglossus millepunctatus lizard is used. The proposed method achieved a maximum efficiency of \(91.11\,\%\).

Real time Kinect evaluation of therapeutical gestures

This paper presents a recognition scheme for therapeutical gestures by using data from a Kinect sensor. Experimental studies are conducted to determine whether the proposed calculation of angles relevant for the definition of therapeutic gestures yields results that can be used to determine, if the movement is being correctly executed. Calculation of the angles using information from two Kinect sensors is performed in order to determine, if it produces a significant improvement in the robustness of the angles used for recognition. The results show that the calculation of the angles formed with the planes of motion is robust enough for a wide variety of therapeutical gestures. They also show that unless the movement is directed towards the sensor the improvement with the use of two Kinects is marginal when calculating the angles of interest.