Diego Campanini

This paper presents the approach of the iHealth-Chile-1 team for the shared task of Large-Scale Radiology Report Generation at the BioNLP workshop, inspired by progress in large multimodal models for processing images and text. In this work, we leverage LLaVA, a Visual-Language Model (VLM), composed of a vision-encoder, a vision-language connector or adapter, and a large language model able to process text and visual embeddings. We achieve our best result by enriching the input prompt of LLaVA with the text output of a simpler report generation model. With this enriched-prompt technique, we improve our results in 4 of 5 metrics (BLEU-4, Rouge-L, BertScore and F1-RadGraph,), only doing in-context learning. Moreover, we provide details about different architecture settings, fine-tuning strategies, and dataset configurations.

This paper presents the approaches of the iHealth-Chile-3 and iHealth-Chile-2 teams for the shared task of Large-Scale Radiology Report Generation at the BioNLP workshop. Inspired by prior work on template-based report generation, both teams focused on exploring various template-based strategies, using predictions from multi-label image classifiers as input. Our best approach achieved a modest F1-RadGraph score of 19.42 on the findings hidden test set, ranking 7th on the leaderboard. Notably, we consistently observed a discrepancy between our classification metrics and the F1-CheXbert metric reported on the leaderboard, which always showed lower scores. This suggests that the F1-CheXbert metric may be missing some of the labels mentioned by the templates.

In this undergraduate thesis, an object detection system (localization and classification) is developed based on convolutional neural networks (CNNs) and two classical machine learning methods: Random Forest (RF) and Support Vector Machines (SVMs). The idea is to improve, using the aforementioned classifiers, the performance of the detection system known as Faster R-CNN (Regions with CNN features). The Faster R-CNN system is based on the region proposal concept to generate candidate samples that may correspond to objects and subsequently produce two outputs: one corresponding to regression, which characterizes object localization, and another corresponding to confidence scores associated with the predicted bounding boxes. Both outputs are generated by fully connected layers. In this work, the output that generates the confidence scores is modified so that, at this point, a classifier (RF or SVM) is connected to generate the system’s output scores. In this way, the goal is to improve the performance of the Faster R-CNN system.

The UChile HomeBreakers team is an effort of the Department of Electrical Engineering of the Universidad de Chile. The team has participated in the RoboCup@ Home league since 2007, and its social robot Bender obtained the@ Home Innovation Award in 2007 and 2008. As a team with strong expertise in robot vision, object recognition, and human-robot interaction, we believe that we can provide interesting features to the league. This year our main research focus is object recognition and its manipulation, because one of the principal abilities of a service robot is the interaction with objects. For this reason the team incorporated a new manipulation system, using the ROS package MoveIT!, and carried out a comparison among object recognition methods. Another important improvement in our social robot is a new face that allows it to more easily display emotions. Additionally, a long-term memory to store non-redundant information about people and objects with which the robot has interacted, as well as places and dates where sessions have been carried out, was implemented.