Abstract: In this paper, we propose an object segmentation framework, based on the popular bag of features (BoF), which can process several images per second while achieving a good segmentation accuracy assigning an object category to every pixel of the image. We propose an efficient color descriptor to complement the information obtained by a typical gradient-based local descriptor. Results show that color proves to be a useful cue to increase the segmentation accuracy, specially in large homogeneous regions. Then, we extend the Hierarchical K-Means codebook using the recently proposed Vector of Locally Aggregated Descriptors method. Finally, we show that the BoF method can be easily parallelized since it is applied locally, thus the time necessary to process an image is further reduced. The performance of the proposed method is evaluated in the standard PASCAL 2007 Segmentation Challenge object segmentation dataset.

Abstract: This paper presents details of a general purpose micro-task on-demand platform based on the crowdsourcing philosophy. This platform was specifically developed for mobile devices in order to exploit the strengths of such devices; namely: i) massivity, ii) ubiquity and iii) embedded sensors. The combined use of mobile platforms and the crowdsourcing model allows to tackle from the simplest to the most complex tasks. Users experience is the highlighted feature of this platform (this fact is extended to both task-proposer and tasksolver). Proper tools according with a specific task are provided to a task-solver in order to perform his/her job in a simpler, faster and appealing way. Moreover, a task can be easily submitted by just selecting predefined templates, which cover a wide range of possible applications. Examples of its usage in computer vision and computer games are provided illustrating the potentiality of the platform.

Abstract: The cost of drawing object bounding boxes (ie labeling) for millions of images is prohibitively high. For instance, labeling pedestrians in a regular urban image could take 35 seconds on average. Active learning aims to reduce the cost of labeling by selecting only those images that are informative to improve the detection network accuracy. In this paper, we propose a method to perform active learning of object detectors based on convolutional neural networks. We propose a new image-level scoring process to rank unlabeled images for their automatic selection, which clearly outperforms classical scores. The proposed method can be applied to videos and sets of still images. In the former case, temporal selection rules can complement our scoring process. As a relevant use case, we extensively study the performance of our method on the task of pedestrian detection. Overall, the experiments show that the proposed method performs better than random selection.

Abstract: Road scene segmentation is important in computer vision for different applications such as autonomous driving and pedestrian detection. Recovering the 3D structure of road scenes provides relevant contextual information to improve their understanding.
In this paper, we use a convolutional neural network based algorithm to learn features from noisy labels to recover the 3D scene layout of a road image. The novelty of the algorithm relies on generating training labels by applying an algorithm trained on a general image dataset to classify on–board images. Further, we propose a novel texture descriptor based on a learned color plane fusion to obtain maximal uniformity in road areas. Finally, acquired (off–line) and current (on–line) information are combined to detect road areas in single images.
From quantitative and qualitative experiments, conducted on publicly available datasets, it is concluded that convolutional neural networks are suitable for learning 3D scene layout from noisy labels and provides a relative improvement of 7% compared to the baseline. Furthermore, combining color planes provides a statistical description of road areas that exhibits maximal uniformity and provides a relative improvement of 8% compared to the baseline. Finally, the improvement is even bigger when acquired and current information from a single image are combined

Abstract: This paper focuses on near infrared (NIR) image colorization by using a Generative Adversarial Network (GAN) architecture model. The proposed architecture consists of two stages. Firstly, it learns to colorize the given input, resulting in a RGB image. Then, in the second stage, a discriminative model is used to estimate the probability that the generated image came from the training dataset, rather than the image automatically generated. The proposed model starts the learning process from scratch, because our set of images is very dierent from the dataset used in existing pre-trained models, so transfer learning strategies cannot be used. Infrared image colorization is an important problem when human perception need to be considered, e.g, in remote sensing applications. Experimental results with a large set of real images are provided showing the validity of the proposed approach.