Mengyi Shan

Mengyi Shan, Brian Curless, Ira Kemelmacher-Shlizerman et al. · uw

We present a system that automatically brings street view imagery to life by populating it with naturally behaving, animated pedestrians and vehicles. Our approach is to remove existing people and vehicles from the input image, insert moving objects with proper scale, angle, motion, and appearance, plan paths and traffic behavior, as well as render the scene with plausible occlusion and shadowing effects. The system achieves these by reconstructing the still image street scene, simulating crowd behavior, and rendering with consistent lighting, visibility, occlusions, and shadows. We demonstrate results on a diverse range of street scenes including regular still images and panoramas.

Zhangsihao Yang, Mingyuan Zhou, Mengyi Shan et al.

Our paper aims to generate diverse and realistic animal motion sequences from textual descriptions, without a large-scale animal text-motion dataset. While the task of text-driven human motion synthesis is already extensively studied and benchmarked, it remains challenging to transfer this success to other skeleton structures with limited data. In this work, we design a model architecture that imitates Generative Pretraining Transformer (GPT), utilizing prior knowledge learned from human data to the animal domain. We jointly train motion autoencoders for both animal and human motions and at the same time optimize through the similarity scores among human motion encoding, animal motion encoding, and text CLIP embedding. Presenting the first solution to this problem, we are able to generate animal motions with high diversity and fidelity, quantitatively and qualitatively outperforming the results of training human motion generation baselines on animal data. Additionally, we introduce AnimalML3D, the first text-animal motion dataset with 1240 animation sequences spanning 36 different animal identities. We hope this dataset would mediate the data scarcity problem in text-driven animal motion generation, providing a new playground for the research community.

Zhangsihao Yang, Mengyi Shan, Mohammad Farazi et al.

Human video generation task has gained significant attention with the advancement of deep generative models. Generating realistic videos with human movements is challenging in nature, due to the intricacies of human body topology and sensitivity to visual artifacts. The extensively studied 2D media generation methods take advantage of massive human media datasets, but struggle with 3D-aware control; whereas 3D avatar-based approaches, while offering more freedom in control, lack photorealism and cannot be harmonized seamlessly with background scene. We propose AMG, a method that combines the 2D photorealism and 3D controllability by conditioning video diffusion models on controlled rendering of 3D avatars. We additionally introduce a novel data processing pipeline that reconstructs and renders human avatar movements from dynamic camera videos. AMG is the first method that enables multi-person diffusion video generation with precise control over camera positions, human motions, and background style. We also demonstrate through extensive evaluation that it outperforms existing human video generation methods conditioned on pose sequences or driving videos in terms of realism and adaptability.

Mengyi Shan, Zecheng He, Haoyu Ma et al.

Can a video generation model be repurposed as an interactive world simulator? We explore the affordance perception potential of text-to-video models by teaching them to predict human-environment interaction. Given a scene image and a prompt describing human actions, we fine-tune the model to insert a person into the scene, while ensuring coherent behavior, appearance, harmonization, and scene affordance. Unlike prior work, we infer human affordance for video generation (i.e., where to insert a person and how they should behave) from a single scene image, without explicit conditions like bounding boxes or body poses. An in-depth study of cross-attention heatmaps demonstrates that we can uncover the inherent affordance perception of a pre-trained video model without labeled affordance datasets.

Mengyi Shan, Brian Curless, Ira Kemelmacher-Shlizerman et al. · uw

We challenge text-to-image models with generating escape room puzzle images that are visually appealing, logically solid, and intellectually stimulating. While base image models struggle with spatial relationships and affordance reasoning, we propose a hierarchical multi-agent framework that decomposes this task into structured stages: functional design, symbolic scene graph reasoning, layout synthesis, and local image editing. Specialized agents collaborate through iterative feedback to ensure the scene is visually coherent and functionally solvable. Experiments show that agent collaboration improves output quality in terms of solvability, shortcut avoidance, and affordance clarity, while maintaining visual quality.

Roy Or-El, Xuan Luo, Mengyi Shan et al.

We introduce a high resolution, 3D-consistent image and shape generation technique which we call StyleSDF. Our method is trained on single-view RGB data only, and stands on the shoulders of StyleGAN2 for image generation, while solving two main challenges in 3D-aware GANs: 1) high-resolution, view-consistent generation of the RGB images, and 2) detailed 3D shape. We achieve this by merging a SDF-based 3D representation with a style-based 2D generator. Our 3D implicit network renders low-resolution feature maps, from which the style-based network generates view-consistent, 1024x1024 images. Notably, our SDF-based 3D modeling defines detailed 3D surfaces, leading to consistent volume rendering. Our method shows higher quality results compared to state of the art in terms of visual and geometric quality.

Mengyi Shan, TJ Tsai

This paper tackles the problem of verifying the authenticity of speech recordings from world leaders. Whereas previous work on detecting deep fake or tampered audio focus on scrutinizing an audio recording in isolation, we instead reframe the problem and focus on cross-verifying a questionable recording against trusted references. We present a method for cross-verifying a speech recording against a reference that consists of two steps: aligning the two recordings and then classifying each query frame as matching or non-matching. We propose a subsequence alignment method based on the Needleman-Wunsch algorithm and show that it significantly outperforms dynamic time warping in handling common tampering operations. We also explore several binary classification models based on LSTM and Transformer architectures to verify content at the frame level. Through extensive experiments on tampered speech recordings of Donald Trump, we show that our system can reliably detect audio tampering operations of different types and durations. Our best model achieves 99.7% accuracy for the alignment task at an error tolerance of 50 ms and a 0.43% equal error rate in classifying audio frames as matching or non-matching.

Mengyi Shan, TJ Tsai

This paper studies the problem of automatically generating piano score following videos given an audio recording and raw sheet music images. Whereas previous works focus on synthetic sheet music where the data has been cleaned and preprocessed, we instead focus on developing a system that can cope with the messiness of raw, unprocessed sheet music PDFs from IMSLP. We investigate how well existing systems cope with real scanned sheet music, filler pages and unrelated pieces or movements, and discontinuities due to jumps and repeats. We find that a significant bottleneck in system performance is handling jumps and repeats correctly. In particular, we find that a previously proposed Jump DTW algorithm does not perform robustly when jump locations are unknown a priori. We propose a novel alignment algorithm called Hierarchical DTW that can handle jumps and repeats even when jump locations are not known. It first performs alignment at the feature level on each sheet music line, and then performs a second alignment at the segment level. By operating at the segment level, it is able to encode domain knowledge about how likely a particular jump is. Through carefully controlled experiments on unprocessed sheet music PDFs from IMSLP, we show that Hierarachical DTW significantly outperforms Jump DTW in handling various types of jumps.

TJ Tsai, Daniel Yang, Mengyi Shan et al.

This article investigates a cross-modal retrieval problem in which a user would like to retrieve a passage of music from a MIDI file by taking a cell phone picture of several lines of sheet music. This problem is challenging for two reasons: it has a significant runtime constraint since it is a user-facing application, and there is very little relevant training data containing cell phone images of sheet music. To solve this problem, we introduce a novel feature representation called a bootleg score which encodes the position of noteheads relative to staff lines in sheet music. The MIDI representation can be converted into a bootleg score using deterministic rules of Western musical notation, and the sheet music image can be converted into a bootleg score using classical computer vision techniques for detecting simple geometrical shapes. Once the MIDI and cell phone image have been converted into bootleg scores, we can estimate the alignment using dynamic programming. The most notable characteristic of our system is that it has no trainable weights at all -- only a set of about 40 hyperparameters. With a training set of just 400 images, we show that our system generalizes well to a much larger set of 1600 test images from 160 unseen musical scores. Our system achieves a test F measure score of 0.89, has an average runtime of 0.90 seconds, and outperforms baseline systems based on music object detection and sheet-audio alignment. We provide extensive experimental validation and analysis of our system.

Daniel Yang, Thitaree Tanprasert, Teerapat Jenrungrot et al.

This paper investigates a cross-modal retrieval problem in which a user would like to retrieve a passage of music from a MIDI file by taking a cell phone picture of a physical page of sheet music. While audio-sheet music retrieval has been explored by a number of works, this scenario is novel in that the query is a cell phone picture rather than a digital scan. To solve this problem, we introduce a mid-level feature representation called a bootleg score which explicitly encodes the rules of Western musical notation. We convert both the MIDI and the sheet music into bootleg scores using deterministic rules of music and classical computer vision techniques for detecting simple geometric shapes. Once the MIDI and cell phone image have been converted into bootleg scores, we estimate the alignment using dynamic programming. The most notable characteristic of our system is that it does test-time adaptation and has no trainable weights at all -- only a set of about 30 hyperparameters. On a dataset containing 1000 cell phone pictures taken of 100 scores of classical piano music, our system achieves an F measure score of .869 and outperforms baseline systems based on commercial optical music recognition software.