Gledson Melotti

  1. Home
  2. Team
  3. Gledson Melotti
Gledson is currently in the PhD in electrical and computers engineering at the University of Coimbra (UC). He is developing research in the area of autonomous vehicle perception systems. He has a master's degree in electrical engineering from the Federal University of Minas Gerais (UFMG-Brazil), a specialization in control and instrumentation engineering at the Faculty of the Centro Leste (UCL-Brazil) and a degree in electrical engineering from the Federal University of São João Del- Rei (UFSJ-Brazil). He has basic knowledge in evolutionary computation applied in PID controls, linear and nonlinear dynamic systems. He is currently a professor at the Federal Institute of Education, Science and Technology of Espírito Santo - São Mateus (Brazil).

Publications

Multimodal Deep-Learning for Object Recognition Combining Camera and LIDAR Data

Object detection and recognition is a key component of autonomous robotic vehicles, as evidenced by the continuous efforts made by the robotic community on areas related to object detection and sensory perception systems. This paper presents a study on multisensor (camera and LIDAR) late fusion strategies for object recognition. In this work, LIDAR data is processed as 3D points and also by means of a 2D representation in the form of depth map (DM), which is obtained by projecting the LIDAR 3D point cloud into a 2D image plane followed by an upsampling strategy which generates a high-resolution 2D range view. A CNN network (Inception V3) is used as classification method on the RGB images, and on the DMs (LIDAR modality). A 3D- network (the PointNet), which directly performs classification on the 3D point clouds, is also considered in the experiments. One of the motivations of this work consists of incorporating the distance to the objects, as measured by the LIDAR, as a relevant cue to improve the classification performance. A new range- based average weighting strategy is proposed, which considers the relationship between the deep-models’ performance and the distance of objects. A classification dataset, based on the KITTI database, is used to evaluate the deep-models, and to support the experimental part. We report extensive results in terms of single modality i.e., using RGB and LIDAR models individually, and late fusion multimodality approaches. (CITATION: Gledson Melotti, Cristiano Premebida, Nuno Gonçalves, "Multimodal Deep-Learning for Object Recognition Combining Camera and LIDAR Data" 2020 IEEE International Conference on Autonomous Robot Systems and Competitions (ICARSC), Ponta Delgada, Portugal, pp. 177-182)

  • Date: 01/10/2020
  • //
  • Featured In: 2020 IEEE International Conference on Autonomous Robot Systems and Competitions (ICARSC), Ponta Delgada, Portugal, pp. 177-182
  • //
  • Publication Type: Conference Papers
  • //
  • Author(s): Gledson Melotti, Cristiano Premebida, Nuno Gonçalves
  • //
  • DOI: 10.1109/ICARSC49921.2020.9096138
  • //
  • Download File
  • //
  • Visit Website

Probabilistic Object Classification using CNN ML-MAP layers

Deep networks are currently the state-of-the-art for sensory perception in autonomous driving and robotics. However, deep models often generate overconfident predictions precluding proper probabilistic interpretation which we argue is due to the nature of the SoftMax layer. To reduce the overconfidence without compromising the classification performance, we introduce a CNN probabilistic approach based on dis- tributions calculated in the network’s Logit layer. The approach enables Bayesian inference by means of ML and MAP layers. Experiments with calibrated and the proposed prediction layers are carried out on object classification using data from the KITTI database. Results are reported for camera (RGB) and LiDAR (range-view) modalities, where the new approach shows promising performance compared to SoftMax.

  • Date: 24/08/2020
  • //
  • Featured In: ECCV Workshop on Perception for Autonomous Driving (PAD)
  • //
  • Publication Type: Conference Papers
  • //
  • Author(s): Gledson Melotti, Cristiano Premebida, Jordan Bird, Diego Faria, and Nuno Gonçalves
  • //
  • Download File
  • //
  • Visit Website

Reducing Overconfidence Predictions in Autonomous Driving Perception

In state-of-the-art deep learning for object recognition, Softmax and Sigmoid layers are most commonly employed as the predictor outputs. Such layers often produce overconfidence predictions rather than proper probabilistic scores, which can thus harm the decision-making of ‘critical’ perception systems applied in autonomous driving and robotics. Given this, we propose a probabilistic approach based on distributions calculated out of the Logit layer scores of pre-trained networks which are then used to constitute new decision layers based on Maximum Likelihood (ML) and Maximum a-Posteriori (MAP) inference. We demonstrate that the hereafter called ML and MAP layers are more suitable for probabilistic interpretations than Softmax and Sigmoid-based predictions for object recognition.We explore distinct sensor modalities via RGB images and LiDARs (RV: range-view) data from the KITTI and Lyft Level-5 datasets, where our approach shows promising performance compared to the usual Softmax and Sigmoid layers, with the benefit of enabling interpretable probabilistic predictions. Another advantage of the approach introduced in this paper is that the so-called ML and MAP layers can be implemented in existing trained networks, that is, the approach benefits from the output of the Logit layer of pre-trained networks. Thus, there is no need to carry out a new training phase since the ML and MAP layers are used in the test/prediction phase. The Classification results are presented using reliability diagrams, while detection results are illustrated using precision-recall curves.

  • Date: 16/05/2022
  • //
  • Featured In: IEEE Access, vol. 10, pp. 54805-54821, 2022
  • //
  • Publication Type: Journal Articles
  • //
  • Author(s): G. Melotti, C. Premebida, J. J. Bird, D. R. Faria and N. Gonçalves
  • //
  • DOI: 10.1109/ACCESS.2022.3175195
  • //
  • Download File
  • //
  • Visit Website

Probabilistic Approach for Road-Users Detection

Object detection in autonomous driving applications implies the detection and tracking of semantic objects that are commonly native to urban driving environments, as pedestrians and vehicles. One of the major challenges in state-of-the-art deep-learning based object detection are false positives which occur with overcon dent scores. This is highly undesirable in autonomous driving and other critical robotic-perception domains because of safety concerns. This paper proposes an approach to alleviate the problem of overcon dent predictions by introducing a novel probabilistic layer to deep object detection networks in testing. The suggested approach avoids the tradi- tional Sigmoid or Softmax prediction layer which often produces overcon dent predictions. It is demonstrated that the proposed technique reduces overcon dence in the false positives without degrading the performance on the true positives. The approach is validated on the 2D-KITTI objection detection through the YOLOV4 and SECOND (Lidar-based detector). The proposed approach enables interpretable probabilistic predictions without the requirement of re-training the network and therefore is very practical.

  • Date: 03/05/2023
  • //
  • Featured In: IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
  • //
  • Publication Type: Journal Articles
  • //
  • Author(s): Gledson Melotti, Weihao Lu, Pedro Conde, Dezong Zhao, Alireza Asvadi, Nuno Gon ̧calves, Cristiano Premebida
  • //
  • DOI: 10.1109/TITS.2023.3268578
  • //
  • Download File
  • //
  • Visit Website