Fernando De la Torre
Research Professor
Studies computer vision and machine learning for human sensing, including facial analysis and generative models.
Fernando De la Torre's lab at CMU works on embodied & active vision and generative & foundation models. 46 papers since 2023, 28 at top venues, mostly ICCV and WACV. 7 PhD students and 29 master's students. 244 alumni. Applications close .
At a glance
- papers, 2023–2026
- 12/ year
- at T1 venues
- 61%
- PhD students
- 7
- MS students
- 29
- alumni listed
- 244
Papers by yearfrom the lab homepage, 2023–2026
- 9 papers2026 · T1 7 · collecting
- 15 papers2025 · T1 6
- 17 papers2024 · T1 10
- 5 papers2023 · T1 5
T1 venuesOtherVenues: ICCV, WACV, ECCV, SIGGRAPH Asia, CVPR
Research focuspaper titles that use each topic's words
- Generative & foundation models7 papers
- Embodied & active vision1 paper
- facial0 papers
- studies computer0 papers
- robotics institute0 papers
Paper list
From the lab homepage.
20269 papers
- CVPR (T1)CGHair: Compact Gaussian Hair Reconstruction with Card Clustering (opens in a new tab)
- ECCV (T1)Diagnosing Aerial-View Detectors using Generative Models (opens in a new tab)
- 3DVFrom Blurry to Believable: Enhancing Low-quality Talking Heads with 3D Generative Priors (opens in a new tab)
- 3DVGarmentCrafter: Progressive Novel View Synthesis for Single-View 3D Garment Reconstruction and Editing (opens in a new tab)
- SIGGRAPH Asia (T1)Lift4D: Harmonizing Single-View 3D Estimation for 4D Reconstruction In-the-Wild (opens in a new tab)
- TACL (T1)LLAMADRS: Evaluating Open-Source LLMs on Real Clinical Interviews—To Reason or Not to Reason? (opens in a new tab)
- CVPR (T1)Multi-view Consistent 3D Gaussian Head Avatars without Multi-view Generation (opens in a new tab)
- SIGGRAPH Asia (T1)OmniFabric: Coherent UV Space Texture Synthesis for 3D Garment Reconstruction (opens in a new tab)
- ECCV (T1)Racing in Volume with Flow Ensembles (opens in a new tab)
202515 papers
- ICCV (T1)Adapting Vehicle Detectors for Aerial Imagery to Unseen Domains with Weak Supervision (opens in a new tab)
- FGCustom Condition Generation for Zero-Shot Human-Scene Interactions Synthesis (opens in a new tab)
- SIGGRAPH (T1)Echoes of the Coliseum: Towards 3D Live streaming of Sports Events (opens in a new tab)
- Expert Systems with ApplicationsExploring Image and Skeleton-Based Action Recognition Approaches for Clinical In-Bed Classification of Simulated Epileptic Seizure Movements (opens in a new tab)
- ICCV (T1)GAS: Generative Avatar Synthesis from a Single Image (opens in a new tab)
- WACVGHOST: Grounded Human Motion Generation with Open Vocabulary Scene-and-Text Contexts (opens in a new tab)
- ICCV (T1)Improving Noise Efficiency in Privacy-preserving Dataset Distillation (opens in a new tab)
- ICCV (T1)LightSwitch: Multi-view Relighting with Material-guided Diffusion (opens in a new tab)
- 3DVMaterialFusion: Enhancing Inverse Rendering with Material Diffusion Priors (opens in a new tab)
- Proceedings of the International Conference on Multimodal Interaction - Workshop (ICMI) WorkshopsMultimodal Fusion with LLMs for Engagement Prediction in Natural Conversation (opens in a new tab)
- FGOASIS: Object-guided Attention for Text-conditional Diffusion Synthesis of Human Interaction Sequences (opens in a new tab)
- Findings of EMNLPOn the Fine-Grained Planning Abilities of VLM Web Agents (opens in a new tab)
202417 papers
- WACVA Sequential Learning-based Approach for Monocular Human Performance Capture (opens in a new tab)
- SIGGRAPH Asia (T1)Consolidating Attention Features for Multi-view Image Editing (opens in a new tab)
- WACVD³GU: Multi-target Active Domain Adaptation via Enhancing Domain Alignment (opens in a new tab)
- Image and Vision ComputingDeep learning methods for single camera based clinical in-bed movement action recognition (opens in a new tab)
- CVPR (T1)Domain Gap Embeddings for Generative Dataset Augmentation (opens in a new tab)
- NeurIPS (T1)Doubly Hierarchical Geometric Representations for Strand-based Human Hairstyle Generation (opens in a new tab)
- WACVExploring the Impact of Rendering Method and Motion Quality on Model Performance when Using a Multi-view Synthetic Data for Action Recognition (opens in a new tab)
- SIGGRAPH Asia (T1)FabricDiffusion: High-Fidelity Texture Transfer for 3D Garments Generation from In-The-Wild Images (opens in a new tab)
- ECCV (T1)FAMOUS: High-Fidelity Monocular 3D Human Digitization Using View Synthesis (opens in a new tab)
- ECCV (T1)Generalizable Human Gaussians (GHG) for Sparse View Synthesis (opens in a new tab)
- NeurIPS (T1)Hamba: Single-view 3D Hand Reconstruction with Graph-guided Bi-Scanning Mamba (opens in a new tab)
- WACVMotionGPT: Human Motion Synthesis with Improved Diversity and Realism via GPT-3 Prompting (opens in a new tab)
20235 papers
- ICCV (T1)A Latent Space of Stochastic Diffusion Models for Zero-Shot Image Editing and Guidance (opens in a new tab)
- ICCV (T1)Data-Free Class-Incremental Hand Gesture Recognition (opens in a new tab)
- ICCV (T1)ITI-GEN: Inclusive Text-to-Image Generation (opens in a new tab)
- ICCV (T1)PATMAT: Person Aware Tuning of Mask Aware Transformer for Face Inpainting (opens in a new tab)
- CVPR (T1)Zero-shot Model Diagnosis (ZOOM) (opens in a new tab)
Where alumni went244 listed
Apply to CMUFall 2027
CMU SCS Computer Science PhD
- Round
- Single round
- Opens
- GRE
- See program page
- English
- TOEFL / IELTS / Duolingo (F-1/J-1 non-native speakers)
- Fee
- USD 100 (USD 80 by early deadline)
Confirmed on the program page.
Similar labsnearest by research summary
- CNZhejiang
Hao Chen
- 3D vision & neural rendering
- GBUCL
Sophia Bano
- Recognition, detection & segmentation
- CNTsinghua
Song-Chun Zhu
- Vision-language & multimodal
- CNZhejiang
Mingli Song
113 T1 - USGeorgia Tech
Zsolt Kira
- Computational imaging & photography
- GBImperial
Krystian Mikolajczyk
- 3D vision & neural rendering
Your notes
Save this lab to keep notes.
Data
Checked against the lab homepage.
- Collected
- Verified
- Papers
- The lab homepage
- Members
- Lab homepage
- Alumni
- Lab homepage
Tiers are venue classifications, not lab rankings.
CMUcloses