This paper presents HunyuanOCR, a commercial-grade, open-source, and lightweight (1B parameters) Vision-Language Model (VLM) dedicated to OCR tasks. The architecture comprises a Native Vision Transformer (ViT) and a lightweight LLM connected via an MLP adapter. HunyuanOCR demonstrates superior performance, outperforming commercial APIs, traditional pipelines, and larger models (e.g., Qwen3-VL-4B). Specifically, it surpasses current public solutions in perception tasks (Text Spotting, Parsing) and excels in semantic tasks (IE, Text Image Translation), securing first place in the ICDAR 2025 DIMT Challenge (Small Model Track). Furthermore, it achieves state-of-the-art (SOTA) results on OCRBench among VLMs with fewer than 3B parameters.
HunyuanOCR achieves breakthroughs in three key aspects: 1) Unifying Versatility and Efficiency: We implement comprehensive support for core capabilities including spotting, parsing, IE, VQA, and translation within a lightweight framework. This addresses the limitations of narrow "OCR expert models" and inefficient "General VLMs". 2) Streamlined End-to-End Architecture: Adopting a pure end-to-end paradigm eliminates dependencies on pre-processing modules (e.g., layout analysis). This fundamentally resolves error propagation common in traditional pipelines and simplifies system deployment. 3) Data-Driven and RL Strategies: We confirm the critical role of high-quality data and, for the first time in the industry, demonstrate that Reinforcement Learning (RL) strategies yield significant performance gains in OCR tasks.
HunyuanOCR is officially open-sourced on HuggingFace. We also provide a high-performance deployment solution based on vLLM, placing its production efficiency in the top tier. We hope this model will advance frontier research and provide a solid foundation for industrial applications.
Ultra Light OCR Competition is a Chinese scene text recognition competition jointly organized by CSIG (China Society of Image and Graphics) and Baidu, Inc. In addition to focusing on common problems in Chinese scene text recognition, such as long text length and massive characters, we need to balance the trade-off of model scale and accuracy since the model size limitation in the competition is 10M. From experiments in aspects of data, model, training, etc, we proposed a general and effective method for Chinese scene text recognition, which got us second place among over 100 teams with accurac…
Sber AI（俄罗斯联邦储蓄银行人工智能研究院）；Sber Devices（俄罗斯联邦储蓄银行设备部门）；Samara National Research University（萨马拉国立研究型大学）；Lomonosov Moscow State University（罗蒙诺索夫莫斯科国立大学）
In the report we propose six new implementations of ruCLIP model trained on our 240M pairs. The accuracy results are compared with original CLIP model with Ru-En translation (OPUS-MT) on 16 datasets from different domains. Our best implementations outperform CLIP + OPUS-MT solution on most of the datasets in few-show and zero-shot tasks. In the report we briefly describe the implementations and concentrate on the conducted experiments. Inference execution time comparison is also presented in the report.
This technical report presents the 2nd winning model for AQTC, a task newly introduced in CVPR 2022 LOng-form VidEo Understanding (LOVEU) challenges. This challenge faces difficulties with multi-step answers, multi-modal, and diverse and changing button representations in video. We address this problem by proposing a new context ground module attention mechanism for more effective feature mapping. In addition, we also perform the analysis over the number of buttons and ablation study of different step networks and video features. As a result, we achieved the overall 2nd place in LOVEU competit…
In this report, we introduce our real-time 2D object detection system for the realistic autonomous driving scenario. Our detector is built on a newly designed YOLO model, called YOLOX. On the Argoverse-HD dataset, our system achieves 41.0 streaming AP, which surpassed second place by 7.8/6.1 on detection-only track/fully track, respectively. Moreover, equipped with TensorRT, our model achieves the 30FPS inference speed with a high-resolution input size (e.g., 1440-2304). Code and models will be available at https://github.com/Megvii-BaseDetection/YOLOX
University of Bonn（波恩大学）；National University of Singapore（新加坡国立大学）
This technical report extends our work presented in [9] with more experiments. In [9], we tackle long-term video understanding, which requires reasoning from current and past or future observations and raises several fundamental questions. How should temporal or sequential relationships be modelled? What temporal extent of information and context needs to be processed? At what temporal scale should they be derived? [9] addresses these questions with a flexible multi-granular temporal aggregation framework. In this report, we conduct further experiments with this framework on different tasks an…
This report presents our submission to the MS COCO Captioning Challenge 2015. The method uses Convolutional Neural Network activations as an embedding to find semantically similar images. From these images, the most typical caption is selected based on unigram frequencies. Although the method received low scores with automated evaluation metrics and in human assessed average correctness, it is competitive in the ratio of captions which pass the Turing test and which are assessed as better or equal to human captions.
This is a brief technical report of our proposed method for Multiple-Object Tracking (MOT) Challenge in Complex Environments. In this paper, we treat the MOT task as a two-stage task including human detection and trajectory matching. Specifically, we designed an improved human detector and associated most of detection to guarantee the integrity of the motion trajectory. We also propose a location-wise matching matrix to obtain more accurate trace matching. Without any model merging, our method achieves 66.672 HOTA and 93.971 MOTA on the DanceTrack challenge dataset.
Georgia Institute of Technology（佐治亚理工学院）；Anonymous University（匿名大学）
In this work, we describe our method for tackling the valence-arousal estimation challenge from ABAW FG-2020 Competition. The competition organizers provide an in-the-wild Aff-Wild2 dataset for participants to analyze affective behavior in real-life settings. We use MIMAMO Net \cite{deng2020mimamo} model to achieve information about micro-motion and macro-motion for improving video emotion recognition and achieve Concordance Correlation Coefficient (CCC) of 0.415 and 0.511 for valence and arousal on the reselected validation set.
Institute of Industrial Science, the University of Tokyo（东京大学生产技术研究所）
In this report, we describe the technical details of our submission to the 2021 EPIC-KITCHENS-100 Unsupervised Domain Adaptation Challenge for Action Recognition. Leveraging multiple modalities has been proved to benefit the Unsupervised Domain Adaptation (UDA) task. In this work, we present Multi-Modal Mutual Enhancement Module (M3EM), a deep module for jointly considering information from multiple modalities to find the most transferable representations across domains. We achieve this by implementing two sub-modules for enhancing each modality using the context of other modalities. The first…
Information Technologies Institute, Centre for Research & Technology（信息技术研究所，研究与技术中心）；School of Medicine, Aristotle University of Thessaloniki（亚里士多德大学医学院）
The objective of Dem@Care is the development of a complete system providing personal health services to people with dementia, as well as medical professionals and caregivers, by using a multitude of sensors, for context-aware, multi-parametric monitoring of lifestyle, ambient environment, and health parameters. Multi-sensor data analysis, combined with intelligent decision making mechanisms, will allow an accurate representation of the person's current status and will provide the appropriate feedback, both to the person and the associated caregivers, enhancing the standard clinical workflow. W…
Image classification has always been a hot and challenging task. This paper is a brief report to our submission to the VIPriors Image Classification Challenge. In this challenge, the difficulty is how to train the model from scratch without any pretrained weight. In our method, several strong backbones and multiple loss functions are used to learn more representative features. To improve the models' generalization and robustness, efficient image augmentation strategies are utilized, like autoaugment and cutmix. Finally, ensemble learning is used to increase the performance of the models. The f…
Université François Rabelais of Tours（图尔弗朗索瓦·拉伯雷大学）
ImageNet is a large scale and publicly available image database. It currently offers more than 14 millions of images, organised according to the WordNet hierarchy. One of the main objective of the creators is to provide to the research community a relevant database for visual recognition applications such as object recognition, image classification or object localisation. However, only a few visual descriptors of the images are available to be used by the researchers. Only SIFT-based features have been extracted from a subset of the collection. This technical report presents the extraction of …
This report describes a 2nd place solution of the detection challenge which is held within CVPR 2020 Retail-Vision workshop. Instead of going further considering previous results this work mainly aims to verify previously observed takeaways by re-experimenting. The reliability and reproducibility of the results are reached by incorporating a popular object detection toolbox - MMDetection. In this report, I firstly represent the results received for Faster-RCNN and RetinaNet models, which were taken for comparison in the original work. Then I describe the experiment results with more advanced m…
In the SSLAD-Track 3B challenge on continual learning, we propose the method of COntinual Learning with Transformer (COLT). We find that transformers suffer less from catastrophic forgetting compared to convolutional neural network. The major principle of our method is to equip the transformer based feature extractor with old knowledge distillation and head expanding strategies to compete catastrophic forgetting. In this report, we first introduce the overall framework of continual learning for object detection. Then, we analyse the key elements' effect on withstanding catastrophic forgetting …
Wangxuan Institute of Computer Technology, Peking University（北京大学王选计算机研究所）
In this technical report, we briefly introduce the solutions of our team `PKU-WICT-MIPL' for the PIC Makeup Temporal Video Grounding (MTVG) Challenge in ACM-MM 2022. Given an untrimmed makeup video and a step query, the MTVG aims to localize a temporal moment of the target makeup step in the video. To tackle this task, we propose a phrase relationship mining framework to exploit the temporal localization relationship relevant to the fine-grained phrase and the whole sentence. Besides, we propose to constrain the localization results of different step sentence queries to not overlap with each o…
In this work, we describe our method for tackling the valence-arousal estimation challenge from ABAW2 ICCV-2021 Competition. The competition organizers provide an in-the-wild Aff-Wild2 dataset for participants to analyze affective behavior in real-life settings. We use a two stream model to learn emotion features from appearance and action respectively. To solve data imbalanced problem, we apply label distribution smoothing (LDS) to re-weight labels. Our proposed method achieves Concordance Correlation Coefficient (CCC) of 0.591 and 0.617 for valence and arousal on the validation set of Aff-wi…
Institute for Infocomm Research, Agency for Science, Technology and Research（新加坡科技研究局资讯通信研究院）；National University of Singapore（新加坡国立大学）
In this report, we present the technical details of our submission to the EPIC-KITCHENS-100 Unsupervised Domain Adaptation (UDA) Challenge for Action Recognition 2022. This task aims to adapt an action recognition model trained on a labeled source domain to an unlabeled target domain. To achieve this goal, we propose an action-aware domain adaptation framework that leverages the prior knowledge induced from the action recognition task during the adaptation. Specifically, we disentangle the source features into action-relevant features and action-irrelevant features using the learned action cla…
Technical Report for ICML 2024 TiFA Workshop MLLM Attack Challenge: Suffix Injection and Projected Gradient Descent Can Easily Fool An MLLM
School of Computing, National University of Singapore（新加坡国立大学计算机学院）；School of Computer Science and Technology, Harbin Institute of Technology (Shenzhen)（哈尔滨工业大学（深圳）计算机科学与技术学院）
This paper is a technical report about our submission for the ECCV 2018 3DRMS Workshop Challenge on Semantic 3D Reconstruction \cite{Tylecek2018rms}. In this paper, we address 3D semantic reconstruction for autonomous navigation using co-learning of depth map and semantic segmentation. The core of our pipeline is a deep multi-task neural network which tightly refines depth and also produces accurate semantic segmentation maps. Its inputs are an image and a raw depth map produced from a pair of images by standard stereo vision. The resulting semantic 3D point clouds are then merged in order to …
Team AcieLee: Technical Report for EPIC-SOUNDS Audio-Based Interaction Recognition Challenge 2023
Southwest University（西南大学）；Samsung Research China-Beijing(SRC-B)（三星中国研究院-北京）；Institute of Computing Technology, Chinese Academy of Sciences（中国科学院计算技术研究所）
