dodo detector | Douglas Meneghetti

During my stay at FEI University Center, I was responsible for integrating object detection techniques into a domestic robot. I worked with libraries such as OpenCV and the TensorFlow Object Detection API as well as ROS, the Robot Operating System. This section provides a centralized collection of resources I’ve created during my years working with computer vision and object detection.

Since you’re here, take a look at the set of Python scripts I provide to help in the creation of TFRecord files and label maps for the TensorFlow Object Detection API, as well as TXT annotation files for YOLOv3. I also authored a brief tutorial on how to train a detection model using the TensorFlow Object Detection API with the help of my scripts.

dodo detector (Python)

dodo detector is a Python package that encapsulates OpenCV object detection via keypoint detection and feature matching, as well as the TensorFlow Object Detection API in a single place. See a simple tutorial here. This code was used in the experiments for the paper (Ferreira et al., 2022).

dodo detector ros (ROS)

I am also the creator of dodo detector ros, a ROS package that allows dodo detector to interface with USB cameras as well as both Kinects v1 and v2, in order to detect objects and place them in a 3D environment using the Kinect point cloud. This code was used to develop Hera (Aquino Junior et al., 2019), a domestic robot that won the 2023 RoboCup, as well as several Latin American robotics competitions.

References

2022

CAPTION: Caption Analysis with Proposed Terms, Image of Objects, and Natural Language Processing

Leonardo Anjoletto Ferreira , Douglas De Rizzo Meneghetti , Paulo Eduardo Santos , and 1 more author

SN Computer Science, Jul 2022

Abs

This paper proposes a novel algorithm, called CAPTION, for identifying and correcting errors in automatically generated image captions. The algorithm combines Deep Learning (DL) for object detection in images with Natural Language Processing techniques. CAPTION has been tested in the following three tasks: (1) classify a caption as correct or not; (2) detect wrong words in the caption, and (3) suggest text corrections. Results show that our method is superior with respect to others evaluated in the same data set in the error correction task. These other methods are generally based exclusively on DL models. This work shows that, although semantics still has not been used at its fullest in this type of task, a combination of DL with Natural Language Processing tools presents a better overall performance than using DL methods alone.

2019

HERA: Home Environment Robot Assistant

Plinio Thomaz Aquino Junior , Bruno de Freitas Vece Perez , Douglas De Rizzo Meneghetti , and 11 more authors

In II Brazilian Humanoid Robot Workshop (BRAHUR) and III Brazilian Workshop on Service Robotics (BRASERO) , Apr 2019