Song Yan

Jinyu Yang, Zhe Li, Song Yan et al.

RGBD object tracking is gaining momentum in computer vision research thanks to the development of depth sensors. Although numerous RGBD trackers have been proposed with promising performance, an in-depth review for comprehensive understanding of this area is lacking. In this paper, we firstly review RGBD object trackers from different perspectives, including RGBD fusion, depth usage, and tracking framework. Then, we summarize the existing datasets and the evaluation metrics. We benchmark a representative set of RGBD trackers, and give detailed analyses based on their performances. Particularly, we are the first to provide depth quality evaluation and analysis of tracking results in depth-friendly scenarios in RGBD tracking. For long-term settings in most RGBD tracking videos, we give an analysis of trackers' performance on handling target disappearance. To enable better understanding of RGBD trackers, we propose robustness evaluation against input perturbations. Finally, we summarize the challenges and provide open directions for this community. All resources are publicly available at https://github.com/memoryunreal/RGBD-tracking-review.

Cheng Li, Ziang Leng, Chenxi Yan et al.

Role-playing chatbots built on large language models have drawn interest, but better techniques are needed to enable mimicking specific fictional characters. We propose an algorithm that controls language models via an improved prompt and memories of the character extracted from scripts. We construct ChatHaruhi, a dataset covering 32 Chinese / English TV / anime characters with over 54k simulated dialogues. Both automatic and human evaluations show our approach improves role-playing ability over baselines. Code and data are available at https://github.com/LC1332/Chat-Haruhi-Suzumiya .

Chenfeng Wang, Wei He, Xuhan Zhu et al.

In language reasoning, longer chains of thought consistently yield better performance, which naturally suggests that visual latent reasoning may likewise benefit from longer latent sequences. However, we discover a counterintuitive phenomenon: the performance of existing latent visual reasoning methods systematically degrades as the latent sequence grows longer. We reveal the root cause: Information Gain Collapse -- autoregressive generation makes each step highly dependent on prior outputs, so subsequent tokens can barely introduce new information. We further identify that heavily pooled ($\geq 128\times$) image embeddings used as supervision targets provide no more signal than meaningless placeholders. Motivated by these insights, we propose SCOLAR (Self-COnsistent LAtent Reasoning), which introduces a lightweight detransformer that leverages the LLM's full-sequence hidden states to generate auxiliary visual tokens in a single shot, with each token independently anchored to the original visual space. Combined with three-stage SFT and ALPO reinforcement learning, SCOLAR extends acceptable latent CoT length by over $30\times$, achieves state-of-the-art among open-source models on real-world reasoning benchmarks (+14.12% over backbone), and demonstrates strong out-of-distribution generalization.

Jiyao Wang, Song Yan, Xiao Yang et al.

Silent automation failures, where a system fails to detect a hazard without warning, pose a critical safety challenge for partially automated vehicles. While research has mostly focused on takeover requests, how to support a driver in silent failure remains underexplored. We conducted a multi-modal driving simulator study with 48 participants to investigate how different Prospective Situation Awareness Enhancement (PSAE) interfaces, delivered via augmented reality head-up display, affect takeover performance. By integrating behavioral, subjective psychological, and physiological data, our analysis suggests that situational awareness (SA) serves as an important moderating factor through which PSAE interfaces improve takeover performance. Further, we found that providing perceptual cues was most effective in enhancing SA, while communicating system intent was superior for building trust. Finally, we identified a potential correlate of SA in the neuroactivity. Overall, this paper contributes to understanding how transparency-oriented interfaces may support drivers and provides design insights into HMI design for silent failures.

Ling Team, Binwei Zeng, Chao Huang et al.

In this technical report, we tackle the challenges of training large-scale Mixture of Experts (MoE) models, focusing on overcoming cost inefficiency and resource limitations prevalent in such systems. To address these issues, we present two differently sized MoE large language models (LLMs), namely Ling-Lite and Ling-Plus (referred to as "Bailing" in Chinese, spelled Bǎilíng in Pinyin). Ling-Lite contains 16.8 billion parameters with 2.75 billion activated parameters, while Ling-Plus boasts 290 billion parameters with 28.8 billion activated parameters. Both models exhibit comparable performance to leading industry benchmarks. This report offers actionable insights to improve the efficiency and accessibility of AI development in resource-constrained settings, promoting more scalable and sustainable technologies. Specifically, to reduce training costs for large-scale MoE models, we propose innovative methods for (1) optimization of model architecture and training processes, (2) refinement of training anomaly handling, and (3) enhancement of model evaluation efficiency. Additionally, leveraging high-quality data generated from knowledge graphs, our models demonstrate superior capabilities in tool use compared to other models. Ultimately, our experimental findings demonstrate that a 300B MoE LLM can be effectively trained on lower-performance devices while achieving comparable performance to models of a similar scale, including dense and MoE models. Compared to high-performance devices, utilizing a lower-specification hardware system during the pre-training phase demonstrates significant cost savings, reducing computing costs by approximately 20%. The models can be accessed at https://huggingface.co/inclusionAI.

Song Yan, Min Li, Bi Xinliang et al.

In text-driven content generation (T2C) diffusion model, semantic of generated content is mostly attributed to the process of text embedding and attention mechanism interaction. The initial noise of the generation process is typically characterized as a random element that contributes to the diversity of the generated content. Contrary to this view, this paper reveals that beneath the random surface of noise lies strong analyzable patterns. Specifically, this paper first conducts a comprehensive analysis of the impact of random noise on the model's generation. We found that noise not only contains rich semantic information, but also allows for the erasure of unwanted semantics from it in an extremely simple way based on information theory, and using the equivalence between the generation process of diffusion model and semantic injection to inject semantics into the cleaned noise. Then, we mathematically decipher these observations and propose a simple but efficient training-free and universal two-step "Semantic Erasure-Injection" process to modulate the initial noise in T2C diffusion model. Experimental results demonstrate that our method is consistently effective across various T2C models based on both DiT and UNet architectures and presents a novel perspective for optimizing the generation of diffusion model, providing a universal tool for consistent generation.

Song Yan, Hui Wei, Jinlong Fei et al.

Various (text) prompt filters and (image) safety checkers have been implemented to mitigate the misuse of Text-to-Image (T2I) models in creating Not-Safe-For-Work (NSFW) content. In order to expose potential security vulnerabilities of such safeguards, multimodal jailbreaks have been studied. However, existing jailbreaks are limited to prompt-specific and image-specific perturbations, which suffer from poor scalability and time-consuming optimization. To address these limitations, we propose Universally Unfiltered and Unseen (U3)-Attack, a multimodal jailbreak attack method against T2I safeguards. Specifically, U3-Attack optimizes an adversarial patch on the image background to universally bypass safety checkers and optimizes a safe paraphrase set from a sensitive word to universally bypass prompt filters while eliminating redundant computations. Extensive experimental results demonstrate the superiority of our U3-Attack on both open-source and commercial T2I models. For example, on the commercial Runway-inpainting model with both prompt filter and safety checker, our U3-Attack achieves $~4\times$ higher success rates than the state-of-the-art multimodal jailbreak attack, MMA-Diffusion.

Song Yan, Jinyu Yang, Jani Käpylä et al.

RGBD (RGB plus depth) object tracking is gaining momentum as RGBD sensors have become popular in many application fields such as robotics.However, the best RGBD trackers are extensions of the state-of-the-art deep RGB trackers. They are trained with RGB data and the depth channel is used as a sidekick for subtleties such as occlusion detection. This can be explained by the fact that there are no sufficiently large RGBD datasets to 1) train deep depth trackers and to 2) challenge RGB trackers with sequences for which the depth cue is essential. This work introduces a new RGBD tracking dataset - Depth-Track - that has twice as many sequences (200) and scene types (40) than in the largest existing dataset, and three times more objects (90). In addition, the average length of the sequences (1473), the number of deformable objects (16) and the number of annotated tracking attributes (15) have been increased. Furthermore, by running the SotA RGB and RGBD trackers on DepthTrack, we propose a new RGBD tracking baseline, namely DeT, which reveals that deep RGBD tracking indeed benefits from genuine training data. The code and dataset is available at https://github.com/xiaozai/DeT

Song Yan, Jinyu Yang, Ales Leonardis et al.

Depth (D) indicates occlusion and is less sensitive to illumination changes, which make depth attractive modality for Visual Object Tracking (VOT). Depth is used in RGBD object tracking where the best trackers are deep RGB trackers with additional heuristic using depth maps. There are two potential reasons for the heuristics: 1) the lack of large RGBD tracking datasets to train deep RGBD trackers and 2) the long-term evaluation protocol of VOT RGBD that benefits from heuristics such as depth-based occlusion detection. In this work, we study how far D-only tracking can go if trained with large amounts of depth data. To compensate the lack of depth data, we generate depth maps for tracking. We train a "Depth-DiMP" from the scratch with the generated data and fine-tune it with the available small RGBD tracking datasets. The depth-only DiMP achieves good accuracy in depth-only tracking and combined with the original RGB DiMP the end-to-end trained RGBD-DiMP outperforms the recent VOT 2020 RGBD winners.

Song Yan, Joni-Kristian Kämäräinen

Accurate estimation of anthropometric body measurements from RGB images has many potential applications in industrial design, online clothing, medical diagnosis and ergonomics. Research on this topic is limited by the fact that there exist only generated datasets which are based on fitting a 3D body mesh to 3D body scans in the commercial CAESAR dataset. For 2D only silhouettes are generated. To circumvent the data bottleneck, we introduce a new 3D scan dataset of 2,675 female and 1,474 male scans. We also introduce a small dataset of 200 RGB images and tape measured ground truth. With the help of the two new datasets we propose a part-based shape model and a deep neural network for estimating anthropometric measurements from 2D images. All data will be made publicly available.

Song Yan, Johan Wirta, Joni-Kristian Kämäräinen

We propose a full processing pipeline to acquire anthropometric measurements from 3D measurements. The first stage of our pipeline is a commercial point cloud scanner. In the second stage, a pre-defined body model is fitted to the captured point cloud. We have generated one male and one female model from the SMPL library. The fitting process is based on non-rigid Iterative Closest Point (ICP) algorithm that minimizes overall energy of point distance and local stiffness energy terms. In the third stage, we measure multiple circumference paths on the fitted model surface and use a non-linear regressor to provide the final estimates of anthropometric measurements. We scanned 194 male and 181 female subjects and the proposed pipeline provides mean absolute errors from 2.5 mm to 16.0 mm depending on the anthropometric measurement.

Yanlin Qian, Song Yan, Joni-Kristian Kämäräinen et al.

In the real world, a scene is usually cast by multiple illuminants and herein we address the problem of spatial illumination estimation. Our solution is based on detecting gray pixels with the help of flash photography. We show that flash photography significantly improves the performance of gray pixel detection without illuminant prior, training data or calibration of the flash. We also introduce a novel flash photography dataset generated from the MIT intrinsic dataset.