美图
3D Vision Algorithm Intern (3D Human Reconstruction & Tracking)
3D Vision Algorithm Intern (3D Human Reconstruction & Tracking)
发布于 大约 10 小时前实习/见习
澳大利亚
无经验要求
实习生
仅现场办公
博士
研究与开发 (研发)
PyTorch
Diffusion Models
Smpl
Aesthetic Enhancement
Human Reconstruction
Multi-View Generation
AI 估算 · 20k–30k
Based on Sydney AI internship rates and company scale, estimated monthly in RMB.
职位详情
关于这个职位
This internship focuses on 3D vision algorithms for human reconstruction and tracking, working with real product-scale datasets to develop aesthetic enhancement and spatiotemporally consistent editing technologies. You will collaborate with Meitu's R&D team in Sydney, contributing to both production and research outcomes.
最低要求
PhD candidate in Computer Science, Artificial Intelligence, Intelligent Arts, or related fields
Proficient in Python, with experience using common libraries (e.g., NumPy, Transformers, Diffusers, Open3D), and strong hands-on experience with deep learning frameworks such as PyTorch
Strong, sustained interest in image/video aesthetics or 3D art; resilient mindset and openness to change
Solid engineering implementation skills and rigorous experimental practices; able to independently complete module-level tasks
Proficient with AI-assisted programming tools; maintains clean, well-structured code
Strong ability to read, analyze, summarize, and reproduce academic papers
Clear logical thinking, strong problem-decomposition skills, and curiosity about underlying principles
Long-term interest in AI + imaging / 3D / aesthetics
Proactive, self-driven, and willing to explore new approaches and challenges
Alignment with a product culture centered on user experience and technical value
工作职责
Research and implement algorithms related to depth estimation, multi-view generation, and 2D / 3D tracking with spatiotemporal reconstruction
Follow state-of-the-art 3D vision papers and open-source projects; reproduce experiments and adapt methods to practical applications
Collaborate with data teams to refine the 3D aesthetic development pipeline, improve data collection and quality evaluation, and establish foundations for high-quality scaling
Explore the integration of human structure priors (Skeleton / SMPL / Mesh) with multi-modal cues such as depth, normals, and optical flow in reconstruction and generative models
Assist in building data processing, evaluation, and visualization tools (e.g., immersive video aesthetic editing) to support rapid iteration
Enable high-quality projection of 3D features into 2D visual outputs, with the goal of producing A-level or above academic publications
优先资格
Prior experience with SMPL, human pose estimation, 3D reconstruction, or tracking
Familiarity with video understanding, diffusion models, autoregressive models, or flow-matching methods
Research projects, competitions, or open-source contributions
Strong mathematical or geometric intuition; excellent aesthetic sensibility and passion for high-quality design
AI 洞察
优缺点分析
优点
- Hands-on experience with real product-scale datasets and cutting-edge algorithms.
- Opportunity to produce both production outcomes and academic publications.
- Work in a collaborative team with strong ties to Meitu's core R&D in China.
- Requires strong self-motivation and ability to work independently on complex problems.
- Fast-paced research environment with need to stay updated with latest papers.
- Internship duration may be limited to 6-12 months, requiring quick ramp-up.
- This role is ideal for PhD students in computer vision or AI who are passionate about 3D human modeling and aesthetic technologies, and who enjoy bridging research and product.
缺点 / 挑战
暂无明显挑战项
角色解读
- Transition to a full-time research engineer or scientist role at Meitu or similar companies.
- Publish high-quality academic papers in top conferences (e.g., CVPR, ICCV).
- Advance to lead projects in 3D vision and generative models for consumer products.
- Research and implement algorithms for 3D human reconstruction and tracking, focusing on depth estimation, multi-view generation, and spatiotemporal consistency.
- Work with product-scale datasets to improve aesthetic enhancement and editing technologies.
- Collaborate with data teams to refine data pipelines and evaluation tools for rapid iteration.
- Strong programming skills in Python and deep learning frameworks like PyTorch.
- Solid understanding of 3D vision fundamentals, including depth estimation, tracking, and reconstruction.
- Ability to read, reproduce, and adapt state-of-the-art research papers.
申请策略
- Tailor your resume to highlight research experience and alignment with human-centric AI technologies.
- Prepare to discuss your approach to problem-decomposition and paper reproduction during interviews.
- Highlight experience with 3D reconstruction, human pose estimation, or SMPL-related projects.
- Showcase any open-source contributions or research publications in 3D vision or generative models.
- Emphasize strong Python and PyTorch skills with examples of implementation.
- Demonstrate interest in aesthetics or imaging through personal projects or portfolios.
- Familiarize yourself with diffusion models, autoregressive models, and flow-matching methods.
- Practice reproducing recent papers from top venues (CVPR, ICCV, NeurIPS) to build implementation skills.
面试指南
- Use STAR method for project-related questions: Situation, Task, Action, Result.
- For implementation questions, discuss step-by-step: data preparation, model architecture, loss functions, evaluation.
- For open-ended design questions, think aloud: clarify requirements, propose options, justify decisions.
- Describe your experience with 3D reconstruction or human pose estimation.
- How would you approach implementing a state-of-the-art paper from scratch?
- Explain the trade-offs between different depth estimation methods (e.g., stereo vs. monocular).
- What is your understanding of SMPL model and its applications?
- How do you stay updated with the latest research in 3D vision?
职位点评
60
综合评分
Top-tier research internship in 3D vision with high growth potential, on-site in Sydney, but compensation is not specified.
更适合这类人
This internship is best suited for candidates who prioritize skill development, research experience, and working with cutting-edge technology over immediate financial compensation.
表现最好
成长发展
相对薄弱
薪资福利
薪资福利40
成长发展90
工作生活50
使命价值60
薪资福利
40较低
The position is an internship with no explicit salary mentioned; typical compensation for AI interns in Sydney is moderate but may not be highly competitive compared to industry rates.
薪资信号未披露（AI估算：20K-30K/月）
成长发展
90较高
Strong emphasis on research and publication, working with state-of-the-art algorithms, and exposure to product-scale datasets offer high developmental value.
技术前沿前沿/新兴技术
技术栈3D Vision、SMPL、Diffusion Models、Depth Estimation、Multi-view Generation
成长机会academic publications、state-of-the-art algorithms、cutting-edge
业务类型ambiguous
工作生活
50较低
The role is on-site in Sydney Central, typical office environment; no mention of flexible work or remote options, and internship duration is fixed.
工作模式仅现场办公
办公地点市区核心地段
加班情况未提及（无法判断）
使命价值
60中等
Meitu's mission of 'Unite Art and Technology' and focus on beauty/AI provides a sense of purpose, but the specific impact on society may be moderate.
行业发展高速增长赛道
社会影响中性/一般
使命信号Unite Art and Technology
创新程度积极采用新技术
Watch Jobs