Kornia

Mentors: Student: Astik Srivastava

This project intends to extend the existing kornia-slam repository to create a visual + inertial measurements based simultaneous localization and mapping pipeline using utilities provided by Kornia-rs. The goal would be to create a VSLAM pipeline compatible with common modalities like monocular, stereo and RGB-D images, with a potential for future GNSS and Lidar integration. This will hopefully provide researchers with a localization and mapping algorithm that can run robustly on low compute hardware.

Mentors: Student: Inchara Jayaram

kornia-rs currently supports only CPU-based computation for image processing operations like resize, warp, and color transforms. This project implements a GPU backend for kornia-tensor and kornia-imgproc, motivated by a concrete Bubbaloop use case: a real-time camera pipeline that needs faster preprocessing to make VLM inference feasible on edge hardware. The work includes a feature-gated GpuAllocator with explicit host-device transfer semantics, allocator-based backend dispatch for tensor operations, and GPU kernels for resize, warp_affine, warp_perspective, and RGB <-> Grayscale conversion using CubeCL.

Mentors: Student: Siddharth Rambhia

This project builds GPU-accelerated image processing kernels for kornia-rs and integrates them into a real-time vision-language model (VLM) pipeline on the Bubbaloop framework, targeting the Jetson Orin Nano. The work has two pillars. First, implementing GPU kernels (resize, normalize, color conversion, cast/layout reordering) using CubeCL, eliminating CPU-GPU synchronization bottlenecks in VLM preprocessing. Second, integrating Qwen2.5-VL-3B via ONNX/TensorRT into a live Bubbaloop camera pipeline, handling dynamic resolution tiling, vision token packing, and memory-constrained inference within the Jetson's 8GB limit. Time permitting, V-JEPA 2 will also be evaluated as a complementary world model for latent scene prediction. The end result positions kornia-rs as a practical foundation for deploying vision-language models on edge devices in Rust.

Mentors: Student: andrew-shc

The aim of this project is to bring State of the Art methods for lightweight Visual Language models into Kornia in Rust. Recently, Hugging Face released a family of small visual language models (SmolVLM) which can be a game changer for the industry to build applications in embedded devices using such AI models.

Mentors: Student: AS1100K

AprilTags, FAST, and ORB are fundamental feature detectors in computer vision and robotics. AprilTags provide robust fiducial markers widely used for calibration, pose estimation, and robot navigation. FAST enables efficient corner detection, which serves as the basis of lightweight real-time applications. ORB combines FAST keypoints with binary descriptors for robust feature extraction and matching in in unstructured environments, making it suitable for tasks such as SLAM and visual localization. This project implements these detectors in Rust, emphasizing efficiency, modularity, and ease of use. All modules are integrated into the kornia-rs ecosystem, expanding its toolkit for robotics, augmented reality, and camera calibration.