VASC Seminar

Virtual Zoom Seminar:  https://cmu.zoom.us/j/92178295543?pwd=L2dwZU5SbDY5NzZZNzZ4ZmFUclRqQT09   Abstract How we represent signals has major implications for the algorithms we build to analyze them. Today, most signals are represented discretely: Images as grids of pixels, shapes as point clouds, audio as grids of amplitudes, etc. If images weren't pixel grids - would we be using convolutional neural networks [...]

Virtual VASC Seminar:  https://cmu.zoom.us/j/92587238250?pwd=S0paYUVBUXozQkFTclMwRUg0MzBNZz09   Abstract: The lens has long been a central element of cameras, since its early use in the mid-nineteenth century by Niepce, Talbot, and Daguerre. The role of the lens, from the Daguerrotype to modern digital cameras, is to refract light to achieve a one-to-one mapping between a point in the scene and a point on the sensor. This effect enables the sensor to compute a particular two-dimensional (2D) [...]

Virtual VASC Seminar: https://cmu.zoom.us/j/96635002737?pwd=RkxGVlJaUTlhcDdGeVBPcnpTS015dz09   Abstract: In this talk, I will show several recent results of my group on learning neural implicit 3D representations, departing from the traditional paradigm of representing 3D shapes explicitly using voxels, point clouds or meshes. Implicit representations have a small memory footprint and allow for modeling arbitrary 3D toplogies at [...]

Virtual VASC - https://cmu.zoom.us/j/99437689110?pwd=cWxuQkIwWlFFZEk0QkVDUVFiN0lTdz09   Abstract: Compositionality is the ability for a model to recognize a concept based on its parts or constituents. This ability is essential to use language effectively as there exists a very large combination of plausible objects, attributes, and actions in the world. We posit that visual recognition models should be [...]

Abstract: Building computer vision systems that understand 3D shape are important for applications including autonomous vehicles, graphics, and VR / AR. If we assume 3D shape supervision, we can now build systems that do a reasonable job at predicting 3D shapes from images. However, 3D supervision is difficult to obtain at scale; therefore we should [...]

Abstract: We live in a 3D world that is dynamic—it is full of life, with inhabitants like people and animals who interact with their environment through moving their bodies. Capturing this complex world in 3D from images has a huge potential for many applications such as compelling mixed reality applications that can interact with people [...]

Abstract: Rapid progress in machine learning, computer vision and graphics leads to successive democratization of media manipulation capabilities. While convincing photo and video manipulation used to require substantial time and skill, modern editors bring (semi-) automated tools that can be used by everyone. Some of the most recent examples include manipulation of human faces, e.g., [...]

Abstract: At Google, advancing the state of the art of computer vision is very impactful as there are billions of users of Google products, many of which require high-quality, artifact-free images. I will share what we learned from successfully launching core computer vision techniques for various Google products, including PhotoScan (Photos), seamless Google Street View [...]

Abstract: Estimating the 6D pose, i.e., 3D rotation and 3D translation, of objects relative to the camera from a single input image has attracted great interest in the computer vision community. Recent works typically address this task by training a deep network to predict the 6D pose given an image as input. While effective on [...]

Abstract: One of the key limitations of deep learning is its inability to generalize to new domains. This talk studies recent attempts at increasing neural network robustness to both natural and adversarial distribution shifts. Robustness to adversarial examples, inputs crafted specifically to fool machine learning models, are arguably the most difficult type of domain shift. [...]