Search

Neural speech codecs have gained great attention for their outstanding reconstruction with discrete token representations. It is a crucial component in generative tasks such as speech coding and large language models (LLM). However, most works based on residual vector quantization perform worse with fewer tokens due to low coding efficiency for modeling complex coupled information. In this paper, we propose a neural speech codec named FreeCodec which employs a more effective encoding framework by decomposing intrinsic properties of speech into different components: 1) a global vector is extracted as the timbre information, 2) a prosody encoder with a long stride level is used to model the prosody information, 3) the content information is from a content encoder. Using different training strategies, FreeCodec achieves state-of-the-art performance in reconstruction and disentanglement scenarios. Results from subjective and objective experiments demonstrate that our framework outperforms existing methods.

This paper presents PCDreamer, a novel method for point cloud completion. Traditional methods typically extract features from partial point clouds to predict missing regions, but the large solution space often leads to unsatisfactory results. More recent approaches have started to use images as extra guidance, effectively improving performance, but obtaining paired data of images and partial point clouds is challenging in practice. To overcome these limitations, we harness the relatively view-consistent multi-view diffusion priors within large models, to generate novel views of the desired shape. The resulting image set encodes both global and local shape cues, which is especially beneficial for shape completion. To fully exploit the priors, we have designed a shape fusion module for producing an initial complete shape from multi-modality input (\ie, images and point clouds), and a follow-up shape consolidation module to obtain the final complete shape by discarding unreliable points introduced by the inconsistency from diffusion priors. Extensive experimental results demonstrate our superior performance, especially in recovering fine details.

Underwater images are often affected by light refraction and absorption, reducing visibility and interfering with subsequent applications. Existing underwater image enhancement methods primarily focus on improving visual quality while overlooking practical implications. To strike a balance between visual quality and application, we propose a heuristic invertible network for underwater perception enhancement, dubbed HUPE, which enhances visual quality and demonstrates flexibility in handling other downstream tasks. Specifically, we introduced an information-preserving reversible transformation with embedded Fourier transform to establish a bidirectional mapping between underwater images and their clear images. Additionally, a heuristic prior is incorporated into the enhancement process to better capture scene information. To further bridge the feature gap between vision-based enhancement images and application-oriented images, a semantic collaborative learning module is applied in the joint optimization process of the visual enhancement task and the downstream task, which guides the proposed enhancement model to extract more task-oriented semantic features while obtaining visually pleasing images. Extensive experiments, both quantitative and qualitative, demonstrate the superiority of our HUPE over state-of-the-art methods. The source code is available at https://github.com/ZengxiZhang/HUPE., Comment: 22 pages, 21 figures

Image super-resolution (SR) is a classical yet still active low-level vision problem that aims to reconstruct high-resolution (HR) images from their low-resolution (LR) counterparts, serving as a key technique for image enhancement. Current approaches to address SR tasks, such as transformer-based and diffusion-based methods, are either dedicated to extracting RGB image features or assuming similar degradation patterns, neglecting the inherent modal disparities between infrared and visible images. When directly applied to infrared image SR tasks, these methods inevitably distort the infrared spectral distribution, compromising the machine perception in downstream tasks. In this work, we emphasize the infrared spectral distribution fidelity and propose a Contourlet refinement gate framework to restore infrared modal-specific features while preserving spectral distribution fidelity. Our approach captures high-pass subbands from multi-scale and multi-directional infrared spectral decomposition to recover infrared-degraded information through a gate architecture. The proposed Spectral Fidelity Loss regularizes the spectral frequency distribution during reconstruction, which ensures the preservation of both high- and low-frequency components and maintains the fidelity of infrared-specific features. We propose a two-stage prompt-learning optimization to guide the model in learning infrared HR characteristics from LR degradation. Extensive experiments demonstrate that our approach outperforms existing image SR models in both visual and perceptual tasks while notably enhancing machine perception in downstream tasks. Our code is available at https://github.com/hey-it-s-me/CoRPLE., Comment: 13 figures, 6 tables

Rapidly developing large language models (LLMs) have brought tremendous intelligent applications. Especially, the GPT-4o's excellent duplex speech interaction ability has brought impressive experience to users. Researchers have recently proposed several multi-modal LLMs in this direction that can achieve user-agent speech-to-speech conversations. This paper proposes a novel speech-text multimodal LLM architecture called Freeze-Omni. Our main contribution is that the speech input and output modalities can be easily connected to a textual LLM while keeping the LLM's parameters frozen throughout the training process. We design a three-stage training strategy for modeling both the speech input and output, enabling Freeze-Omni to obtain speech-to-speech conversation ability using text-speech paired data (such as ASR and TTS data) and only 60,000 multi-round text Q&A data on 8 GPUs. Moreover, we can effectively ensure that the intelligence of the Freeze-Omni in the speech modality is at the same level compared with that in the text modality of its backbone LLM, while achieving low latency end-to-end spoken response. In addition, we also designed a method to achieve duplex dialogue ability through multi-task training, giving Freeze-Omni a more natural style of dialogue ability between users and agents. In summary, Freeze-Omni holds great potential to conduct speech-to-speech dialogue based on a multimodal LLM under the condition of a frozen LLM, avoiding the catastrophic forgetting problem caused by limited data and training resources., Comment: Project Page: https://freeze-omni.github.io/

During the dielectric breakdown process of thin solid-state nanopores, the application of high voltages may cause the formation of multi-nanopores on one chip, which number and sizes are important for their applications. Here, simulations were conducted to mimic the investigation of in situ nanopore detection with scanning ion conductance microscopy (SICM). Results show that SICM can provide accurate nanopore location and relative pore size. Detection resolution is influenced by the dimensions of the applied probe and separation between the probe and membranes, which can be enhanced under large voltages or a concentration gradient., Comment: 22 pages, 6 figuers

Bipolar nanopores, with asymmetric charge distributions, can induce significant ionic current rectification (ICR) at ultra-short lengths, finding potential applications in nanofluidic devices, energy conversion, and other related fields. Here, with simulations, we investigated the characteristics of ion transport and modulation of ICR inside bipolar nanopores. With bipolar nanopores of half-positive and half-negative surfaces, the most significant ICR phenomenon appears at various concentrations. In these cases, ICR ratios are independent of electrolyte types. In other cases where nanopores have oppositely charged surfaces in different lengths, ICR ratios are related to the mobility of anions and cations. The pore length and surface charge density can enhance ICR. As the pore length increases, ICR ratios first increase and then approach their saturation which is determined by the surface charge density. External surface charges of nanopores can promote the ICR phenomenon mainly due to the enhancement of ion enrichment inside nanopores by external surface conductance. The effective width of exterior charged surfaces under various conditions is also explored, which is inversely proportional to the pore length and salt concentration, and linearly related to the pore diameter, surface charge density, and applied voltage. Our results may provide guidance for the design of bipolar porous membranes., Comment: 31 pages, 7 figures

As commonly used implicit geometry representations, the signed distance function (SDF) is limited to modeling watertight shapes, while the unsigned distance function (UDF) is capable of representing various surfaces. However, its inherent theoretical shortcoming, i.e., the non-differentiability at the zero level set, would result in sub-optimal reconstruction quality. In this paper, we propose the scaled-squared distance function (S$^{2}$DF), a novel implicit surface representation for modeling arbitrary surface types. S$^{2}$DF does not distinguish between inside and outside regions while effectively addressing the non-differentiability issue of UDF at the zero level set. We demonstrate that S$^{2}$DF satisfies a second-order partial differential equation of Monge-Ampere-type, allowing us to develop a learning pipeline that leverages a novel Monge-Ampere regularization to directly learn S$^{2}$DF from raw unoriented point clouds without supervision from ground-truth S$^{2}$DF values. Extensive experiments across multiple datasets show that our method significantly outperforms state-of-the-art supervised approaches that require ground-truth surface information as supervision for training. The code will be publicly available at https://github.com/chuanxiang-yang/S2DF.

Audio-LLM introduces audio modality into a large language model (LLM) to enable a powerful LLM to recognize, understand, and generate audio. However, during speech recognition in noisy environments, we observed the presence of illusions and repetition issues in audio-LLM, leading to substitution and insertion errors. This paper proposes a transcription prompt-based audio-LLM by introducing an ASR expert as a transcription tokenizer and a hybrid Autoregressive (AR) Non-autoregressive (NAR) decoding approach to solve the above problems. Experiments on 10k-hour WenetSpeech Mandarin corpus show that our approach decreases 12.2% and 9.6% CER relatively on Test_Net and Test_Meeting evaluation sets compared with baseline. Notably, we reduce the decoding repetition rate on the evaluation set to zero, showing that the decoding repetition problem has been solved fundamentally.

The remarkable multimodal capabilities and interactive experience of GPT-4o underscore their necessity in practical applications, yet open-source models rarely excel in both areas. In this paper, we introduce VITA, the first-ever open-source Multimodal Large Language Model (MLLM) adept at simultaneous processing and analysis of Video, Image, Text, and Audio modalities, and meanwhile has an advanced multimodal interactive experience. Starting from Mixtral 8x7B as a language foundation, we expand its Chinese vocabulary followed by bilingual instruction tuning. We further endow the language model with visual and audio capabilities through two-stage multi-task learning of multimodal alignment and instruction tuning. VITA demonstrates robust foundational capabilities of multilingual, vision, and audio understanding, as evidenced by its strong performance across a range of both unimodal and multimodal benchmarks. Beyond foundational capabilities, we have made considerable progress in enhancing the natural multimodal human-computer interaction experience. VITA is the first step for the open-source community to explore the seamless integration of multimodal understanding and interaction. While there is still lots of work to be done on VITA to get close to close-source counterparts, we hope that its role as a pioneer can serve as a cornerstone for subsequent research. Project Page: https://vita-home.github.io., Comment: Project Page: https://vita-home.github.io

Topologically nontrivial Haldane phase is theoretically proposed to be realized in the 1/3-magnetization ($M$) plateau of spin-1/2 trimer systems. However, the spin excitation gap, typical characteristic of Haldane phase, is not yet experimentally verified. Here, we report the nuclear magnetic resonance investigations into the low-energy spin dynamics in the $S=1/2$ spin-trimer antiferromagnetic chain compound Na$_2$Cu$_3$Ge$_{4-x}$Si$_{x}$O$_{12}$ ($x=0, 0.1\sim1.5$). In the parent compound ($x=0$), the spin-lattice relaxation rate (1/$T_1$) shows significantly different temperature dependence when the external magnetic field is increased above the critical field of $\mu_0$$H_{c}$ = 29 T. The spin excitation gap is evidenced from the thermally activated behavior of $1/T_1(T)$ in the 1/3-$M$ plateau state. By substituting Ge$^{4+}$ with Si$^{4+}$, the critical field for the 1/3-$M$ plateau significantly decreases, e.g. $\mu_0H_{c}=17$ T in $x=1.0$ samples, which results from the suppressed inter-trimer coupling $J_2$. The gapped spin excitation is confirmed again above 17 T, whose size shows temperature-dependent behavior for $\mu_0H\geq25.72$ T. These observations provide further insights into the Haldane physics., Comment: 6 pages, 4 figures

Ionic current rectification (ICR) of charged conical nanopores has various applications in fields including nanofluidics, bio-sensing, and energy conversion, whose function is closely related to the dynamic response of nanopores. The occurrence of ICR originates from the ion enrichment and depletion in conical pores, whose formation is found to be affected by the scanning rate of voltages. Here, through time-dependent simulations, we investigate the variation of ion current under electric fields and the dynamic formation of ion enrichment and depletion, which can reflect the response time of conical nanopores. The response time of nanopores when ion enrichment forms i.e. at the on state is significantly longer than that with the formation of ion depletion i.e. at the off state. Our simulation results reveal the regulation of response time by different nanopore parameters including the surface charge density, pore length, tip, and base radius, as well as the applied conditions such as the voltage and bulk concentration. The response time of nanopores is closely related to the surface charge density, pore length, voltage, and bulk concentration. Our uncovered dynamic response mechanism of the ionic current can guide the design of nanofluidic devices with conical nanopores, including memristors, ionic switches, and rectifiers., Comment: 30 pages, 5 figures

Localizing text in low-light environments is challenging due to visual degradations. Although a straightforward solution involves a two-stage pipeline with low-light image enhancement (LLE) as the initial step followed by detector, LLE is primarily designed for human vision instead of machine and can accumulate errors. In this work, we propose an efficient and effective single-stage approach for localizing text in dark that circumvents the need for LLE. We introduce a constrained learning module as an auxiliary mechanism during the training stage of the text detector. This module is designed to guide the text detector in preserving textual spatial features amidst feature map resizing, thus minimizing the loss of spatial information in texts under low-light visual degradations. Specifically, we incorporate spatial reconstruction and spatial semantic constraints within this module to ensure the text detector acquires essential positional and contextual range knowledge. Our approach enhances the original text detector's ability to identify text's local topological features using a dynamic snake feature pyramid network and adopts a bottom-up contour shaping strategy with a novel rectangular accumulation technique for accurate delineation of streamlined text features. In addition, we present a comprehensive low-light dataset for arbitrary-shaped text, encompassing diverse scenes and languages. Notably, our method achieves state-of-the-art results on this low-light dataset and exhibits comparable performance on standard normal light datasets. The code and dataset will be released.

We performed 23Na NMR and magnetization measurements on an S = 1, quasi-2D honeycomb lattice antiferromagnet Na3Ni2BiO6. A large positive Curie-Weiss constant of 22.9 K is observed. The NMR spectra at low fields are consistent with a "zigzag" magnetic order, indicating a large easy-axis anisotropy. With field applied along the c* axis, the NMR spectra confirm the existence of a 1/3-magnetization plateau phase between 5.1 T and 7.1 T. The transition from the zigzag order to the 1/3-magnetization plateau phase is also found to be a first-order type. A monotonic decrease of the spin gap is revealed in the 1/3-magnetization plateau phase, which reaches zero at a quantum critical field Hc = 8.35 T before entering the fully polarized phase. These data suggest the existence of exchange frustration in the system along with strong ferromagnetic interactions, hosting the possibility for Kitaev physics. Besides, well below the ordered phase, the 1/T1 at high fields shows either a level off or an enhancement upon cooling below 3 K, which suggests the existence of low-energy fluctuations., Comment: 7 pages, 7 figures

With the resistive-pulse technique (RPT), nanopores serve as the nanofluidic sensors of various analytes for their many physical and chemical properties. Here, we focus on the size measurement and its theoretical prediction for sub-200 nm nanoparticles with RPT. Through systematical investigation of the current blockade of nanoparticles across cylindrical nanopores with simulations, Maxwell method considering the shape coefficient and access resistances agrees well with simulation results. However, the widely used integration method of the resistance has distinct deviations in various cases. With the introduction of a correction factor \b{eta} to the integration method, our revised equations can provide good predictions for simulation results. \b{eta} shows a strong dependence on the diameter ratio (d over D) of the nanoparticle and nanopore. Following the same strategy, modified equations are provided for the accurate size prediction for nanoparticles across conical nanopores, where the integration method is the default convenient way. The correction factor \b{eta}' relates to \b{eta} in cylindrical nanopores. \b{eta}' exhibits independence on the pore geometry parameters and diameters of nanoparticles, but dependence on the surface charge density of conical nanopores. Our improved equations can provide theoretical predictions for the accurate size detection of 100-200 nm diameter nanoparticles across cylindrical and conical nanopores., Comment: 29 pages 7 figures

Short nanopores find extensive applications capitalizing on their high throughput and detection resolution. Ionic behaviors through long nanopores are mainly determined by charged inner-pore walls. When pore lengths decrease to sub-200 nm, charged exterior surfaces provide considerable modulation to ion current. We find that the charge status of inner-pore walls affects the modulation of ion current from charged exterior surfaces. For 50-nm-long nanopores with neutral inner-pore walls, charged exterior surfaces on the voltage (surfaceV) and ground (surfaceG) sides enhance and inhibit ion transport by forming ion enrichment and depletion zones inside nanopores, respectively. For nanopores with both charged inner-pore and exterior surfaces, continuous electric double layers enhance ion transport through nanopores significantly. The charged surfaceV results in higher ion current by simultaneously weakening ion depletion at pore entrances and enhancing the intra-pore ion enrichment. The charged surfaceG expedites the exit of ions from nanopores, resulting in a decrease in ion enrichment at pore exits. Through adjustment in the width of charged-ring regions near pore boundaries, the effective charged width of the charged exterior is explored at ~20nm. Our results may provide a theoretical guide for further optimizing the performance of nanopore-based applications, like seawater desalination, biosensing, and osmotic energy conversion., Comment: 18 pages, 5 figures

Short nanopores have various applications in biosensing, desalination, and energy conversion. Here, the modulation of charged exterior surfaces on ionic transport is investigated through simulations with sub-200 nm long nanopores under applied voltages. Detailed analysis of ionic current, electric field strength, and fluid flow inside and outside nanopores reveals that charged exterior surfaces can increase ionic conductance by increasing both the concentration and migration speed of charge carriers. The electric double layers near charged exterior surfaces provide an ion pool and an additional passageway for counterions, which lead to enhanced exterior surface conductance and ionic concentrations at pore entrances and inside the nanopore. We also report that charges on the membrane surfaces increase electric field strengths inside nanopores. The effective width of a ring with surface charges placed at pore entrances (Lcs) is considered as well by studying the dependence of the current on Lcs. We find a linear relationship between the effective Lcs and the surface charge density and voltage, and an inverse relationship between the geometrical pore length and salt concentration. Our results elucidate the modulation mechanism of charged exterior surfaces on ionic transport through short nanopores, which is important for the design and fabrication of porous membranes., Comment: 27 pages, 6 figures

Continuum simulations become an important tool to uncover the mysteries in nanofluidic experiments. However, fluid flow in simulation models is usually unconsidered. Here, systematical simulations are conducted to provide a quantitative understanding of influences from electroosmotic flow (EOF) on ionic transport through nanopores by both types of models with and without consideration of EOF. In nanopores of less than ~10 nm in diameter, counterions dominate ionic current which is always promoted obviously by the convective effect of EOF. In the diameter range from ~10 to ~30 nm, strong EOF induces ion concentration polarization or ion depletion inside nanopores which causes significant decreases in ionic current. For nanopores larger than ~30 nm, due to convective promotion and inhibition of EOF on the transport of counterions and anions, considerable nanopore selectivity to counterions maintains in cases with EOF. Though the difference in total current between both cases decreases with further pore size increasing, the difference in cation/anion current is still considerable. From our results under various pore parameters and applied conditions, the fluid flow should be considered in the simulation cases when EOF is strong. Our work may provide useful guidance for simulation conductance., Comment: 25 pages, 7 figures

Besides the dominant NaCl, natural seawater/river water contains trace multivalent ions, which can provide effective screening to surface charges. Here, in both negatively and positively charged nanopores, influences from divalent ions as counterions and coions have been investigated on the performance of osmotic energy conversion (OEC) under natural salt gradients. As counterions, trace Ca2+ ions can suppress the electric power and conversion efficiency significantly. The reduced OEC performance is due to the bivalence and low diffusion coefficient of Ca2 ions, instead of the uphill transport of divalent ions discovered in the previous work. Effectively screened charged surfaces by Ca2+ ions induce enhanced diffusion of Cl ions which simultaneously decreases the net ion penetration and ionic selectivity of the nanopore. While as coions, Ca2+ ions have weak effects on the OEC performance. The promotion from charged exterior surfaces on OEC processes for ultra-short nanopores is also studied, which effective region is ~200 nm in width beyond pore boundaries independent of the presence of Ca2+ ions. Our results shed light on the physical details of the nanofluidic OEC process under natural seawater/river water conditions, which can provide a useful guide for high-performance osmotic energy harvesting., Comment: 24 pages, 5 figures

As an important property of porous membranes, the surface charge property determines many ionic behaviors of nanopores, such as ionic conductance and selectivity. Based on the dependence of electric double layers on bulk concentrations, ionic conductance through nanopores at high and low concentrations is governed by the bulk conductance and surface charge density, respectively. Here, through the investigation of ionic conductance inside track-etched single polyethylene terephthalate (PET) nanopores under various concentrations, the surface charge density of PET membranes is extracted as around 0.021 C per m2 at pH 10 over measurements with 40 PET nanopores. Simulations show that surface roughness can cause underestimation in surface charge density due to the inhibited electroosmotic flow. Then, the averaged pore size and porosity of track-etched multipore PET membranes are characterized by the developed ionic conductance method. Through coupled theoretical predictions in ionic conductance under high and low concentrations, the averaged pore size and porosity of porous membranes can be obtained simultaneously. Our method provides a simple and precise way to characterize the pore size and porosity of multipore membranes, especially for those with sub-100 nm pores and low porosities., Comment: 25 pages, 4 figures

Fast adapting to unknown peers (partners or opponents) with different strategies is a key challenge in multi-agent games. To do so, it is crucial for the agent to probe and identify the peer's strategy efficiently, as this is the prerequisite for carrying out the best response in adaptation. However, exploring the strategies of unknown peers is difficult, especially when the games are partially observable and have a long horizon. In this paper, we propose a peer identification reward, which rewards the learning agent based on how well it can identify the behavior pattern of the peer over the historical context, such as the observation over multiple episodes. This reward motivates the agent to learn a context-aware policy for effective exploration and fast adaptation, i.e., to actively seek and collect informative feedback from peers when uncertain about their policies and to exploit the context to perform the best response when confident. We evaluate our method on diverse testbeds that involve competitive (Kuhn Poker), cooperative (PO-Overcooked), or mixed (Predator-Prey-W) games with peer agents. We demonstrate that our method induces more active exploration behavior, achieving faster adaptation and better outcomes than existing methods., Comment: ICML 2024. 20 pages

The escalating quality of video generated by advanced video generation methods results in new security challenges, while there have been few relevant research efforts: 1) There is no open-source dataset for generated video detection, 2) No generated video detection method has been proposed so far. To this end, we propose an open-source dataset and a detection method for generated video for the first time. First, we propose a scalable dataset consisting of 964 prompts, covering various forgery targets, scenes, behaviors, and actions, as well as various generation models with different architectures and generation methods, including the most popular commercial models like OpenAI's Sora and Google's Veo. Second, we found via probing experiments that spatial artifact-based detectors lack generalizability. Hence, we propose a simple yet effective \textbf{de}tection model based on \textbf{f}rame \textbf{co}nsistency (\textbf{DeCoF}), which focuses on temporal artifacts by eliminating the impact of spatial artifacts during feature learning. Extensive experiments demonstrate the efficacy of DeCoF in detecting videos generated by unseen video generation models and confirm its powerful generalizability across several commercially proprietary models. Our code and dataset will be released at \url{https://github.com/wuwuwuyue/DeCoF}.

With the rapid progression of deep learning technologies, multi-modality image fusion has become increasingly prevalent in object detection tasks. Despite its popularity, the inherent disparities in how different sources depict scene content make fusion a challenging problem. Current fusion methodologies identify shared characteristics between the two modalities and integrate them within this shared domain using either iterative optimization or deep learning architectures, which often neglect the intricate semantic relationships between modalities, resulting in a superficial understanding of inter-modal connections and, consequently, suboptimal fusion outcomes. To address this, we introduce a text-guided multi-modality image fusion method that leverages the high-level semantics from textual descriptions to integrate semantics from infrared and visible images. This method capitalizes on the complementary characteristics of diverse modalities, bolstering both the accuracy and robustness of object detection. The codebook is utilized to enhance a streamlined and concise depiction of the fused intra- and inter-domain dynamics, fine-tuned for optimal performance in detection tasks. We present a bilevel optimization strategy that establishes a nexus between the joint problem of fusion and detection, optimizing both processes concurrently. Furthermore, we introduce the first dataset of paired infrared and visible images accompanied by text prompts, paving the way for future research. Extensive experiments on several datasets demonstrate that our method not only produces visually superior fusion results but also achieves a higher detection mAP over existing methods, achieving state-of-the-art results., Comment: 10 pages, 12 figures, 3 tables, conference

The current GAN inversion methods typically can only edit the appearance and shape of a single object and background while overlooking spatial information. In this work, we propose a 3D editing framework, 3D-GOI, to enable multifaceted editing of affine information (scale, translation, and rotation) on multiple objects. 3D-GOI realizes the complex editing function by inverting the abundance of attribute codes (object shape/appearance/scale/rotation/translation, background shape/appearance, and camera pose) controlled by GIRAFFE, a renowned 3D GAN. Accurately inverting all the codes is challenging, 3D-GOI solves this challenge following three main steps. First, we segment the objects and the background in a multi-object image. Second, we use a custom Neural Inversion Encoder to obtain coarse codes of each object. Finally, we use a round-robin optimization algorithm to get precise codes to reconstruct the image. To the best of our knowledge, 3D-GOI is the first framework to enable multifaceted editing on multiple objects. Both qualitative and quantitative experiments demonstrate that 3D-GOI holds immense potential for flexible, multifaceted editing in complex multi-object scenes.Our project and code are released at https://3d-goi.github.io .