A5001320062- Advanced Vision and Imaging
- Video Surveillance and Tracking Methods
- Robotics and Sensor-Based Localization
- Domain Adaptation and Few-Shot Learning
- Multimodal Machine Learning Applications
- Human Pose and Action Recognition
- Advanced Image and Video Retrieval Techniques
- 3D Shape Modeling and Analysis
- Fire Detection and Safety Systems
- Landslides and related hazards
- COVID-19 diagnosis using AI
- Anomaly Detection Techniques and Applications
- Image Processing and 3D Reconstruction
École Polytechnique Fédérale de Lausanne
2015-2018
The performance of a classifier trained on data coming from specific domain typically degrades when applied to related but different one. While annotating many samples the new would address this issue, it is often too expensive or impractical. Domain Adaptation has therefore emerged as solution problem; It leverages annotated source domain, in which abundant, train operate target either sparse even lacking altogether. In context, recent trend consists learning deep architectures whose...
We propose an approach for detecting flying objects such as Unmanned Aerial Vehicles (UAVs) and aircrafts when they occupy a small portion of the field view, possibly moving against complex backgrounds, are filmed by camera that itself moves. argue solving difficult problem requires combining both appearance motion cues. To this end we regression-based object-centric stabilization image patches allows us to achieve effective classification on spatio-temporal cubes outperform state-of-the-art...
We propose an efficient approach to exploiting motion information from consecutive frames of a video sequence recover the 3D pose people. Previous approaches typically compute candidate poses in individual and then link them post-processing step resolve ambiguities. By contrast, we directly regress spatio-temporal volume bounding boxes central frame. further show that, for this achieve its full potential, it is essential compensate so that subject remains centered. This allows us effectively...
We propose an approach to detect flying objects such as UAVs and aircrafts when they occupy a small portion of the field view, possibly moving against complex backgrounds, are filmed by camera that itself moves.
We present Kaolin, a PyTorch library aiming to accelerate 3D deep learning research. Kaolin provides efficient implementations of differentiable modules for use in systems. With functionality load and preprocess several popular datasets, native functions manipulate meshes, pointclouds, signed distance functions, voxel grids, mitigates the need write wasteful boilerplate code. packages together graphics including rendering, lighting, shading, view warping. also supports an array loss...
The goal of Deep Domain Adaptation is to make it possible use Nets trained in one domain where there enough annotated training data another little or none. Most current approaches have focused on learning feature representations that are invariant the changes occur when going from other, which means using same network parameters both domains. While some recent algorithms explicitly model by adapting parameters, they either severely restrict changes, significantly increase number parameters....
We propose an approach for on-line detection of small Unmanned Aerial Vehicles (UAVs) and estimation their relative positions velocities in the 3D environment from a single moving camera context sense avoid systems. This problem is challenging both point view, as there are no markers on targets available, tracking perspective, due to misdetection false positives. Furthermore, methods need be computationally light, despite complexity computer vision algorithms, used UAVs with limited payload....
We propose a new method to estimate the 6-dof trajectory of flying object such as quadrotor UAV within 3D airspace monitored using multiple fixed ground cameras. It is based on structure from motion formulation for reconstruction single moving point with known dynamics. Our main contribution bundle adjustment procedure, which in addition optimizing camera poses, regularizes prior dynamics (or specifically flight dynamics). Furthermore, we can infer underlying control input sent UAVs...
Unmanned Aerial Vehicles are becoming increasingly popular for a broad variety of tasks ranging from aerial imagery to objects delivery. With the expansion areas, where drones can be efficiently used, collision risk with other flying increases. Avoiding such collisions would relatively easy task, if all aircrafts in neighboring airspace could communicate each and share their location information. However, it is often case that either information unavailable (e.g. GPS-denied environments) or...
We propose an efficient approach to exploiting motion information from consecutive frames of a video sequence recover the 3D pose people. Previous approaches typically compute candidate poses in individual and then link them post-processing step resolve ambiguities. By contrast, we directly regress spatio-temporal volume bounding boxes central frame. further show that, for this achieve its full potential, it is essential compensate so that subject remains centered. This allows us effectively...
The goal of Deep Domain Adaptation is to make it possible use Nets trained in one domain where there enough annotated training data another little or none. Most current approaches have focused on learning feature representations that are invariant the changes occur when going from other, which means using same network parameters both domains. While some recent algorithms explicitly model by adapting parameters, they either severely restrict changes, significantly increase number parameters....