This paper proposes an automatic method for detection and mapping of Espeletia plants from aerial images acquired by UAV drone. The proposed approach integrated a computer vision for automatic extraction of training zones and tested on three well-established machine learning algorithms to detect regions belonging to Espeletia plants. The main components of the method are: (i) data capture and preprocessing; (ii) automatic extraction of training zones; and (iii) classification procedure using machine learning algorithms. Experimental results show that the method can achieve accurate detection and mapping of Espeletia plants, with up to 98.3% accuracy.

This paper proposes a method for automatic 3D point cloud segmentation of leaves on individual potato (Solanum tubero-sum) plants from DAV-based images. Conventional methods based on visual inspection require significant field work and human effort. The proposed approach uses remote sensing techniques as well as the Voxel Cloud Connectivity Segmentation (VCCS) technique to identify geometrical features and calculate structural attributes. The main component of the method is the SuperVoxel-based leaves segmentation from the tridimensional photogrametric point cloud. The approach is tested using DAV data from an experimental plot. Super-voxel segments are compared with a manual segmentation conducted by experts. The results show that the proposed method allows an accurate leaves segmentation, i.e. it is as good as conventional methods, but much more economic and timely.

Even though convolutional neural networks have become the method of choice in many fields of computer vision, they still lack interpretability and are usually designed manually in a cumbersome trial-and-error process. This paper aims at overcoming those limitations by proposing a deep neural network, which is designed in a systematic fashion and is interpretable, by integrating multiresolution analysis at the core of the deep neural network design. By using the lifting scheme, it is possible to generate a wavelet representation and design a network capable of learning wavelet coefficients in an end-to-end form. Compared to state-of-the-art architectures, the proposed model requires less hyper-parameter tuning and achieves competitive accuracy in image classification tasks. The Code implemented for this research is available at https://github.com/mxbastidasr/DAWN_WACV2020.

In this study, the authors propose a new method for transmission tower detection that involves the use of visual features and the linear content of the scene. For this process, they developed a descriptor based on a grid of two‐dimensional feature descriptors that is useful not only for object detection, but also for tracking the area of interest. For the detection and classification, they used a support vector machine. The experiments were conducted with a dataset of real world images from transmission tower videos that were used to validate the strategy by comparing it with the ground truth. The results showed that the obtained method is fast and appropriate for tower detection in video sequences of environments that include rural and urban areas. The detection took less than 50 ms and was faster than other methods.

A methodology for 3D modeling of virtualized reality objects using neural computing is presented. In this paper the objects are represented in virtualized reality and their 3D data are acquired by one of three acquisition systems: endoneurosonographic equipment (ENS), stereo vision system and non-contact 3D digitizer. These objects are modeled by one of three neural architectures: Multilayer Feed-Forward Neural Network (MLFFNN), Self-Organizing Maps (SOM) and Neural Gas Network (NGN). The 3D virtualized representations correspond to several objects as phantom brain tumors, faces, archaeological items, fruits, among others. We carry out comparisons in terms of computational cost, architectural complexity, training method, training epochs and performance. Finally, we present the modeling results and conclude that SOM and NGN models achieve the best performances and the lowest displaying times, while MLFFNN models have the lowest memory requirements and acceptable training times.

In this paper we present a voice command and mouth gesture based robot command interface which is capable of controlling three degrees of freedom. The gesture set was designed in order to avoid head rotation and translation, and thus relying solely in mouth movements. Mouth segmentation is performed by using the normalized a* component, as in J. Gomez, et al., (October 2008). The gesture detection process is carried out by a Gaussian mixture model (GMM) based classifier. After that, a state machine stabilizes the system response by restricting the number of possible movements depending on the initial state. Voice commands are modeled using a hidden Markov model (HMM) isolated word recognition scheme. The interface was designed taking into account the specific pose restrictions found in the DaVinci assisted surgery command console.

We present an integrated interface that takes advantage of several VR technologies for the exploration of small artifacts in a Museum. This interface allows visitors to observe pieces in 3D from several viewpoints, touch them, feel their weight, and hear their sound at touch. From one hand, it will allow visitors to observe artifacts closer, in order to know more about selected pieces and to enhance their overall experience in a Museum. From the other hand, it leverages existing technologies in order to provide a multimodal interface, which can be easily replaced for others depending on costs or functionality. We describe some of the early results and experiences we provide in this setup.

The Gold Museum, in Bogota, Colombia, displays the largest collection of pre-Hispanic gold artifacts in the world and it has been renovated recently. With funds from the Colombian Government, we have created a multimodal experience that allows visitors to touch, hear, and see small artifacts. Here we present a description of this demo, its functionality, and technical requirements.

We present the iterative development and initial evaluation of a multi-modal platform for interacting with precious small artifacts from the Gold Museum in Bogota. By using a commercial haptic interface, loud speakers, and stereo displays, one can allow visitors to touch, hear, and observe in stereo those precious artifacts. We use this multi-modal interface in a novel way and in a novel context in order to provide virtual replicas that can be weighed, cleaned, and explored as if they were close to a visitor's hand. This platform is currently open to the public, and some of the lessons learned are reported in terms of usability in a real-world museum application.

We propose an algorithm to produce a 3-D CAD model from a set of range data, based on non-uniform rational B-splines (NURBS) surface fitting technique. Our goal is to construct continuous geometric models, assuming that the topology of surface is unknown. In our approach, the triangulated surface is partitioned in quadrilateral patches, using Morse theory. The quadrilateral obtained mesh is regularized by means of the use of geodesic curves and B-splines to obtain a new adequate grid on which to draw NURBS surfaces. Such NURBS surfaces are optimized by means of evolutionary strategies. Further, the patches are smoothly joined guaranteeing continuity C 1 .

In this paper, recurrent (Hopfield-type) neural network associative memories are synthesized, using the perceptron algorithm, in order to locate coffee beans on a tree branch, which is a very important task for harvest automation. The respective training is done on the basis of a set of pictures. The procedure is as follows: first the picture is enhanced in order to adequate the image for the learning process; then, the image is divided in sub-images, from where relevant information is selected as memory patterns. After the training is carried out, an evaluation is performed.

We propose an automated inspection system of manufactured parts using a cloud of 3D measured points of a part provided by a range sensor, and its CAD model. Inspection consists in verifying the accuracy of a part related to a given set of tolerances. It is thus necessary that the 3D measurements be accurate. In the 3D capture of a part, several sources of error can alter the measured values. So, we have to find and model the most influential parameters affecting the accuracy of the range sensor in the digitalization process. This model is used to produce a sensing plan to acquire accurately the geometry of a part. By using the noise model, we introduce a dispersion value for each 3D point acquired. This value of dispersion is shown as a weight factor in the inspection results.