Trained with an unprecedented scale of data, large language models (LLMs) like ChatGPT and GPT-4 exhibit the emergence of significant reasoning abilities from model scaling. Such a trend underscored the potential of training LLMs with unlimited language data, advancing the development of a universal embodied agent. In this work, we introduce the NavGPT, a purely LLM-based instruction-following navigation agent, to reveal the reasoning capability of GPT models in complex embodied scenes by performing zero-shot sequential action prediction for vision-and-language navigation (VLN). At each step, NavGPT takes the textual descriptions of visual observations, navigation history, and future explorable directions as inputs to reason the agent's current status, and makes the decision to approach the target. Through comprehensive experiments, we demonstrate NavGPT can explicitly perform high-level planning for navigation, including decomposing instruction into sub-goal, integrating commonsense knowledge relevant to navigation task resolution, identifying landmarks from observed scenes, tracking navigation progress, and adapting to exceptions with plan adjustment. Furthermore, we show that LLMs is capable of generating high-quality navigational instructions from observations and actions along a path, as well as drawing accurate top-down metric trajectory given the agent's navigation history. Despite the performance of using NavGPT to zero-shot R2R tasks still falling short of trained models, we suggest adapting multi-modality inputs for LLMs to use as visual navigation agents and applying the explicit reasoning of LLMs to benefit learning-based models.
In this paper, we seek to design a report generation model that is able to generate reasonable reports even given different images of various body parts. We start by directly merging multiple datasets and training a single report generation model on this one. We, however, observe that the reports generated in such a simple way only obtain comparable performance compared with that trained separately on each specific dataset. We suspect that this is caused by the dilemma between the diversity of body parts and the limited availability of medical data. To develop robust and generalizable models, it is important to consider a diverse range of body parts and medical conditions. However, collecting a sufficiently large dataset for each specific body part can be difficult due to various factors, such as data availability and privacy concerns. Thus, rather than striving for more data, we propose a single-for-multiple (S4M) framework, which seeks to facilitate the learning of the report generation model with two auxiliary priors: an explicit prior (\ie, feeding radiology-informed knowledge) and an implicit prior (\ie, guided by cross-modal features). Specifically, based on the conventional encoder-decoder report generation framework, we incorporate two extra branches: a Radiology-informed Knowledge Aggregation (RadKA) branch and an Implicit Prior Guidance (IPG) branch. We conduct the experiments on our merged dataset which consists of a public dataset (\ie, IU-Xray) and five private datasets, covering six body parts: chest, abdomen, knee, hip, wrist and shoulder. Our S4M model outperforms all the baselines, regardless of whether they are trained on separate or merged datasets. Code is available at: \url{https://github.com/YtongXie/S4M}.
Deep learning-based approaches have achieved remarkable performance in single-image denoising. However, training denoising models typically requires a large amount of data, which can be difficult to obtain in real-world scenarios. Furthermore, synthetic noise used in the past has often produced significant differences compared to real-world noise due to the complexity of the latter and the poor modeling ability of noise distributions of Generative Adversarial Network (GAN) models, resulting in residual noise and artifacts within denoising models. To address these challenges, we propose a novel method for synthesizing realistic noise using diffusion models. This approach enables us to generate large amounts of high-quality data for training denoising models by controlling camera settings to simulate different environmental conditions and employing guided multi-scale content information to ensure that our method is more capable of generating real noise with multi-frequency spatial correlations. In particular, we design an inversion mechanism for the setting, which extends our method to more public datasets without setting information. Based on the noise dataset we synthesized, we have conducted sufficient experiments on multiple benchmarks, and experimental results demonstrate that our method outperforms state-of-the-art methods on multiple benchmarks and metrics, demonstrating its effectiveness in synthesizing realistic noise for training denoising models.
Volume data is commonly found in many scientific disciplines, like medicine, physics, and biology. Experts rely on robust scientific visualization techniques to extract valuable insights from the data. Recent years have shown path tracing to be the preferred approach for volumetric rendering, given its high levels of realism. However, real-time volumetric path tracing often suffers from stochastic noise and long convergence times, limiting interactive exploration. In this paper, we present a novel method to enable real-time global illumination for volume data visualization. We develop Photon Field Networks -- a phase-function-aware, multi-light neural representation of indirect volumetric global illumination. The fields are trained on multi-phase photon caches that we compute a priori. Training can be done within seconds, after which the fields can be used in various rendering tasks. To showcase their potential, we develop a custom neural path tracer, with which our photon fields achieve interactive framerates even on large datasets. We conduct in-depth evaluations of the method's performance, including visual quality, stochastic noise, inference and rendering speeds, and accuracy regarding illumination and phase function awareness. Results are compared to ray marching, path tracing and photon mapping. Our findings show that Photon Field Networks can faithfully represent indirect global illumination across the phase spectrum while exhibiting less stochastic noise and rendering at a significantly faster rate than traditional methods.
Efficient deep learning-based approaches have achieved remarkable performance in single image super-resolution. However, recent studies on efficient super-resolution have mainly focused on reducing the number of parameters and floating-point operations through various network designs. Although these methods can decrease the number of parameters and floating-point operations, they may not necessarily reduce actual running time. To address this issue, we propose a novel multi-stage lightweight network boosting method, which can enable lightweight networks to achieve outstanding performance. Specifically, we leverage enhanced high-resolution output as additional supervision to improve the learning ability of lightweight student networks. Upon convergence of the student network, we further simplify our network structure to a more lightweight level using reparameterization techniques and iterative network pruning. Meanwhile, we adopt an effective lightweight network training strategy that combines multi-anchor distillation and progressive learning, enabling the lightweight network to achieve outstanding performance. Ultimately, our proposed method achieves the fastest inference time among all participants in the NTIRE 2023 efficient super-resolution challenge while maintaining competitive super-resolution performance. Additionally, extensive experiments are conducted to demonstrate the effectiveness of the proposed components. The results show that our approach achieves comparable performance in representative dataset DIV2K, both qualitatively and quantitatively, with faster inference and fewer number of network parameters.
Implicit Neural Representations (INRs) have recently exhibited immense potential in the field of scientific visualization for both data generation and visualization tasks. However, these representations often consist of large multi-layer perceptrons (MLPs), necessitating millions of operations for a single forward pass, consequently hindering interactive visual exploration. While reducing the size of the MLPs and employing efficient parametric encoding schemes can alleviate this issue, it compromises generalizability for unseen parameters, rendering it unsuitable for tasks such as temporal super-resolution. In this paper, we introduce HyperINR, a novel hypernetwork architecture capable of directly predicting the weights for a compact INR. By harnessing an ensemble of multiresolution hash encoding units in unison, the resulting INR attains state-of-the-art inference performance (up to 100x higher inference bandwidth) and can support interactive photo-realistic volume visualization. Additionally, by incorporating knowledge distillation, exceptional data and visualization generation quality is achieved, making our method valuable for real-time parameter exploration. We validate the effectiveness of the HyperINR architecture through a comprehensive ablation study. We showcase the versatility of HyperINR across three distinct scientific domains: novel view synthesis, temporal super-resolution of volume data, and volume rendering with dynamic global shadows. By simultaneously achieving efficiency and generalizability, HyperINR paves the way for applying INR in a wider array of scientific visualization applications.
In situ visualization and steering of computational modeling can be effectively achieved using reactive programming, which leverages temporal abstraction and data caching mechanisms to create dynamic workflows. However, implementing a temporal cache for large-scale simulations can be challenging. Implicit neural networks have proven effective in compressing large volume data. However, their application to distributed data has yet to be fully explored. In this work, we develop an implicit neural representation for distributed volume data and incorporate it into the DIVA reactive programming system. This implementation enables us to build an in situ temporal caching system with a capacity 100 times larger than previously achieved. We integrate our implementation into the Ascent infrastructure and evaluate its performance using real-world simulations.
The ability to use inductive reasoning to extract general rules from multiple observations is a vital indicator of intelligence. As humans, we use this ability to not only interpret the world around us, but also to predict the outcomes of the various interactions we experience. Generalising over multiple observations is a task that has historically presented difficulties for machines to grasp, especially when requiring computer vision. In this paper, we propose a model that can extract general rules from video demonstrations by simultaneously performing summarisation and translation. Our approach differs from prior works by framing the problem as a multi-sequence-to-sequence task, wherein summarisation is learnt by the model. This allows our model to utilise edge cases that would otherwise be suppressed or discarded by traditional summarisation techniques. Additionally, we show that our approach can handle noisy specifications without the need for additional filtering methods. We evaluate our model by synthesising programs from video demonstrations in the Vizdoom environment achieving state-of-the-art results with a relative increase of 11.75% program accuracy on prior works
Given a piece of text, a video clip and a reference audio, the movie dubbing (also known as visual voice clone V2C) task aims to generate speeches that match the speaker's emotion presented in the video using the desired speaker voice as reference. V2C is more challenging than conventional text-to-speech tasks as it additionally requires the generated speech to exactly match the varying emotions and speaking speed presented in the video. Unlike previous works, we propose a novel movie dubbing architecture to tackle these problems via hierarchical prosody modelling, which bridges the visual information to corresponding speech prosody from three aspects: lip, face, and scene. Specifically, we align lip movement to the speech duration, and convey facial expression to speech energy and pitch via attention mechanism based on valence and arousal representations inspired by recent psychology findings. Moreover, we design an emotion booster to capture the atmosphere from global video scenes. All these embeddings together are used to generate mel-spectrogram and then convert to speech waves via existing vocoder. Extensive experimental results on the Chem and V2C benchmark datasets demonstrate the favorable performance of the proposed method. The source code and trained models will be released to the public.
Invariant learning methods try to find an invariant predictor across several environments and have become popular in OOD generalization. However, in situations where environments do not naturally exist in the data, they have to be decided by practitioners manually. Environment partitioning, which splits the whole training dataset into environments by algorithms, will significantly influence the performance of invariant learning and has been left undiscussed. A good environment partitioning method can bring invariant learning to applications with more general settings and improve its performance. We propose to split the dataset into several environments by finding low-correlated data subsets. Theoretical interpretations and algorithm details are both introduced in the paper. Through experiments on both synthetic and real data, we show that our Decorr method can achieve outstanding performance, while some other partitioning methods may lead to bad, even below-ERM results using the same training scheme of IRM.