ZINA: Multimodal Fine-grained Hallucination Detection and Editing

Multimodal Large Language Models (MLLMs) often generate hallucinations, where the output deviates from the visual content. Given that these hallucinations can take diverse forms, detecting hallucinations at a fine-grained level is essential for comprehensive evaluation and analysis. To this end, we propose a novel task of multimodal fine-grained hallucination detection and editing for MLLMs. Moreover, we propose ZINA, a novel method that identifies hallucinated spans at a fine-grained level, classifies their error types into six categories, and suggests appropriate refinements. To train and evaluate models for this task, we constructed VisionHall, a dataset comprising 6.9k outputs from twelve MLLMs manually annotated by 211 annotators, and 20k synthetic samples generated using a graph-based method that captures dependencies among error types. We demonstrated that ZINA outperformed existing methods, including GPT-4o and LLama-3.2, in both detection and editing tasks.

Mobile Manipulation Instruction Generation from Multiple Images with Automatic Metric Enhancement

We consider the problem of generating free-form mobile manipulation instructions based on a target object image and receptacle image. Conventional image captioning models are not able to generate appropriate instructions because their architectures are typically optimized for single-image. In this study, we propose a model that handles both the target object and receptacle to generate free-form instruction sentences for mobile manipulation tasks. Moreover, we introduce a novel training method that effectively incorporates the scores from both learning-based and n-gram based automatic evaluation metrics as rewards. This method enables the model to learn the co-occurrence relationships between words and appropriate paraphrases. Results demonstrate that our proposed method outperforms baseline methods including representative multimodal large language models on standard automatic evaluation metrics. Moreover, physical experiments reveal that using our method to augment data on language instructions improves the performance of an existing multimodal language understanding model for mobile manipulation.

Future Success Prediction in Open-Vocabulary Object Manipulation Tasks Based on End-Effector Trajectories

This study addresses a task designed to predict the future success or failure of open-vocabulary object manipulation. In this task, the model is required to make predictions based on natural language instructions, egocentric view images before manipulation, and the given end-effector trajectories. Conventional methods typically perform success prediction only after the manipulation is executed, limiting their efficiency in executing the entire task sequence. We propose a novel approach that enables the prediction of success or failure by aligning the given trajectories and images with natural language instructions. We introduce Trajectory Encoder to apply learnable weighting to the input trajectories, allowing the model to consider temporal dynamics and interactions between objects and the end effector, improving the model's ability to predict manipulation outcomes accurately. We constructed a dataset based on the RT-1 dataset, a large-scale benchmark for open-vocabulary object manipulation tasks, to evaluate our method. The experimental results show that our method achieved a higher prediction accuracy than baseline approaches.

Task Success Prediction and Open-Vocabulary Object Manipulation

This study addresses a task designed to predict the future success or failure of open-vocabulary object manipulation. In this task, the model is required to make predictions based on natural language instructions, egocentric view images before manipulation, and the given end-effector trajectories. Conventional methods typically perform success prediction only after the manipulation is executed, limiting their efficiency in executing the entire task sequence. We propose a novel approach that enables the prediction of success or failure by aligning the given trajectories and images with natural language instructions. We introduce Trajectory Encoder to apply learnable weighting to the input trajectories, allowing the model to consider temporal dynamics and interactions between objects and the end effector, improving the model's ability to predict manipulation outcomes accurately. We constructed a dataset based on the RT-1 dataset, a large-scale benchmark for open-vocabulary object manipulation tasks, to evaluate our method. The experimental results show that our method achieved a higher prediction accuracy than baseline approaches.

Open-Vocabulary Mobile Manipulation Based on Double Relaxed Contrastive Learning with Dense Labeling

Growing labor shortages are increasing the demand for domestic service robots (DSRs) to assist in various settings. In this study, we develop a DSR that transports everyday objects to specified pieces of furniture based on open-vocabulary instructions. Our approach focuses on retrieving images of target objects and receptacles from pre-collected images of indoor environments. For example, given an instruction "Please get the right red towel hanging on the metal towel rack and put it in the white washing machine on the left," the DSR is expected to carry the red towel to the washing machine based on the retrieved images. This is challenging because the correct images should be retrieved from thousands of collected images, which may include many images of similar towels and appliances. To address this, we propose RelaX-Former, which learns diverse and robust representations from among positive, unlabeled positive, and negative samples. We evaluated RelaX-Former on a dataset containing real-world indoor images and human annotated instructions including complex referring expressions. The experimental results demonstrate that RelaX-Former outperformed existing baseline models across standard image retrieval metrics. Moreover, we performed physical experiments using a DSR to evaluate the performance of our approach in a zero-shot transfer setting. The experiments involved the DSR to carry objects to specific receptacles based on open-vocabulary instructions, achieving an overall success rate of 75%.

Task Success Prediction for Open-Vocabulary Manipulation Based on Multi-Level Aligned Representations

In this study, we consider the problem of predicting task success for open-vocabulary manipulation by a manipulator, based on instruction sentences and egocentric images before and after manipulation. Conventional approaches, including multimodal large language models (MLLMs), often fail to appropriately understand detailed characteristics of objects and/or subtle changes in the position of objects. We propose Contrastive $\lambda$-Repformer, which predicts task success for table-top manipulation tasks by aligning images with instruction sentences. Our method integrates the following three key types of features into a multi-level aligned representation: features that preserve local image information; features aligned with natural language; and features structured through natural language. This allows the model to focus on important changes by looking at the differences in the representation between two images. We evaluate Contrastive $\lambda$-Repformer on a dataset based on a large-scale standard dataset, the RT-1 dataset, and on a physical robot platform. The results show that our approach outperformed existing approaches including MLLMs. Our best model achieved an improvement of 8.66 points in accuracy compared to the representative MLLM-based model.

DENEB: A Hallucination-Robust Automatic Evaluation Metric for Image Captioning

In this work, we address the challenge of developing automatic evaluation metrics for image captioning, with a particular focus on robustness against hallucinations. Existing metrics are often inadequate for handling hallucinations, primarily due to their limited ability to compare candidate captions with multifaceted reference captions. To address this shortcoming, we propose DENEB, a novel supervised automatic evaluation metric specifically robust against hallucinations. DENEB incorporates the Sim-Vec Transformer, a mechanism that processes multiple references simultaneously, thereby efficiently capturing the similarity between an image, a candidate caption, and reference captions. To train DENEB, we construct the diverse and balanced Nebula dataset comprising 32,978 images, paired with human judgments provided by 805 annotators. We demonstrated that DENEB achieves state-of-the-art performance among existing LLM-free metrics on the FOIL, Composite, Flickr8K-Expert, Flickr8K-CF, Nebula, and PASCAL-50S datasets, validating its effectiveness and robustness against hallucinations.

DM2RM: Dual-Mode Multimodal Ranking for Target Objects and Receptacles Based on Open-Vocabulary Instructions

In this study, we aim to develop a domestic service robot (DSR) that, guided by open-vocabulary instructions, can carry everyday objects to the specified pieces of furniture. Few existing methods handle mobile manipulation tasks with open-vocabulary instructions in the image retrieval setting, and most do not identify both the target objects and the receptacles. We propose the Dual-Mode Multimodal Ranking model (DM2RM), which enables images of both the target objects and receptacles to be retrieved using a single model based on multimodal foundation models. We introduce a switching mechanism that leverages a mode token and phrase identification via a large language model to switch the embedding space based on the prediction target. To evaluate the DM2RM, we construct a novel dataset including real-world images collected from hundreds of building-scale environments and crowd-sourced instructions with referring expressions. The evaluation results show that the proposed DM2RM outperforms previous approaches in terms of standard metrics in image retrieval settings. Furthermore, we demonstrate the application of the DM2RM on a standardized real-world DSR platform including fetch-and-carry actions, where it achieves a task success rate of 82% despite the zero-shot transfer setting. Demonstration videos, code, and more materials are available at https://kkrr10.github.io/dm2rm/.

Nearest Neighbor Future Captioning: Generating Descriptions for Possible Collisions in Object Placement Tasks

Domestic service robots (DSRs) that support people in everyday environments have been widely investigated. However, their ability to predict and describe future risks resulting from their own actions remains insufficient. In this study, we focus on the linguistic explainability of DSRs. Most existing methods do not explicitly model the region of possible collisions; thus, they do not properly generate descriptions of these regions. In this paper, we propose the Nearest Neighbor Future Captioning Model that introduces the Nearest Neighbor Language Model for future captioning of possible collisions, which enhances the model output with a nearest neighbors retrieval mechanism. Furthermore, we introduce the Collision Attention Module that attends regions of possible collisions, which enables our model to generate descriptions that adequately reflect the objects associated with possible collisions. To validate our method, we constructed a new dataset containing samples of collisions that can occur when a DSR places an object in a simulation environment. The experimental results demonstrated that our method outperformed baseline methods, based on the standard metrics. In particular, on CIDEr-D, the baseline method obtained 25.09 points, whereas our method obtained 33.08 points.

Layer-Wise Relevance Propagation with Conservation Property for ResNet

The transparent formulation of explanation methods is essential for elucidating the predictions of neural networks, which are typically black-box models. Layer-wise Relevance Propagation (LRP) is a well-established method that transparently traces the flow of a model's prediction backward through its architecture by backpropagating relevance scores. However, the conventional LRP does not fully consider the existence of skip connections, and thus its application to the widely used ResNet architecture has not been thoroughly explored. In this study, we extend LRP to ResNet models by introducing Relevance Splitting at points where the output from a skip connection converges with that from a residual block. Our formulation guarantees the conservation property throughout the process, thereby preserving the integrity of the generated explanations. To evaluate the effectiveness of our approach, we conduct experiments on ImageNet and the Caltech-UCSD Birds-200-2011 dataset. Our method achieves superior performance to that of baseline methods on standard evaluation metrics such as the Insertion-Deletion score while maintaining its conservation property. We will release our code for further research at https://5ei74r0.github.io/lrp-for-resnet.page/