Pavlo Molchanov

Pavlo Molchanov

Principal Research Scientist and Research Lead

NVIDIA Research

Biography

I am a principal research scientist and research lead with NVIDIA Research since 2015. My research is focused on efficient deep learning and human-centric computer vision in LPR team lead by Jan Kautz. In the area of network efficiency I am interested in methods for model acceleration, inversion, novel architectures and adaptive/conditional inference. In the area of human-centric vision I am working on face/body/hand landmarks and pose estimation, action/gesture recognition and designing novel human-computer interaction systems.

I obtained PhD from Tampere University of Technology, Finland in the area of signal processing in 2014 supervised by Karen Eguiazarian. My dissertation was focused on designing automatic target recognition systems for radars and general radar signal processing. During PhD studies I was honored to receive EuRAD best paper award in 2011 and EuRAD young engineer award in 2013. I received my master degree from National Aerospace University, Kharkiv, Ukraine in Radio Systems focusing on high-order spectrum techniques for signal processing and mentored by Alexander Totsky.

We always look for promising interns and full time positions, feel free to ping me for details. I am also interested in connecting with people who share similar research interests.

Publications

Full list with filter

Gradvit: Gradient inversion of vision transformers. (2022). In CVPR 2022.

PDF Cite Code CVPR2022

Do Gradient Inversion Attacks Make Federated Learning Unsafe?. (2022). arXiv preprint arXiv:2202.06924.

PDF Cite

DRaCoN--Differentiable Rasterization Conditioned Neural Radiance Fields for Articulated Avatars. (2022). arXiv preprint arXiv:2203.15798.

Cite

AViT: Adaptive Tokens for Efficient Vision Transformer. (2021). In CVPR 2022.

PDF Cite Code CVPR2022 (Oral)

GLAMR: Global Occlusion-Aware Human Mesh Recovery with Dynamic Cameras. (2021). In CVPR 2022.

PDF Cite Code Project Video CVPR2022 (Oral)

When to Prune? A Policy towards Early Structural Pruning. (2021). In CVPR 2022.

PDF Cite Project CVPR2022

LANA: Latency Aware Network Acceleration. (2021). In Arxiv.

PDF Cite

Deep Neural Networks are Surprisingly Reversible: A Baseline for Zero-Shot Inversion. (2021). In XAI4CV 2022.

PDF Cite Poster

Adversarial motion modelling helps semi-supervised hand pose estimation. (2021). arXiv preprint arXiv:2106.05954.

PDF Cite

DexYCB: A benchmark for capturing hand grasping of objects. (2021). Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.

PDF Cite Code Project Video