Faster decoding method speeds instance segmentation on gpus
Accelerated Decoding of Centroid Positional Encoding for Instance Segmentation
Computer Vision and Pattern Recognition
Summary
Instance segmentation means finding and outlining every object in a picture, which takes a lot of computing power. The authors focus on the slow step after a model predicts raw data: turning those predictions into usable object outlines, which is usually done inefficiently. They studied a recent way of encoding object positions and created a faster method using graphics cards that work in parallel. Their solution makes the whole process quicker, helping real-time systems work better.
What this means in practice
- •For computer vision engineers: Reduce inference time in real-time instance segmentation models by using an optimized GPU decoding method for centroid positional embeddings.
- •For augmented reality developers: Implement efficient segmentation decoding to improve speed and responsiveness of AR systems that identify objects in live camera feeds.
Authors
Carmelo Scribano, Filippo Muzzini, Nedyalko Prisadnikov, Mohammad Mahdi, Yuqian Fu, Giorgia Franchini, Danda Pani Paudel, Marko Bertogna, Luc Van Gool
Abstract
Beyond model inference, the decoding stage, which converts raw network outputs into task-level representations, constitutes a significant portion of the execution cost. Despite its practical impact, prediction decoding has received comparatively little attention and is often implemented using generic CPU routines or inefficient GPU kernels, limiting the benefits of advances in model efficiency. In this work, we investigate the decoding overhead associated with a recent sinusoidal centroid encoding for Instance Segmentation, in which each pixel regresses a positional embedding of its instance centroid. This approach allows flexible segmentation without predefined proposals, but extracting instance masks from dense embeddings incurs a high computational cost. We present an optimized CUDA-based implementation of the decoding algorithm tailored to this encoding, explicitly addressing challenges related to parallelization, synchronization, and memory access on modern GPUs. Our solution significantly reduces decoding overhead and improves End-to-End inference latency, outperforming both CPU-based approaches and naive GPU implementations. The results demonstrate that efficient decoding is essential to fully exploit the advantages of advanced output representations and highlight the importance of jointly designing encoding schemes and their decoding algorithms for real-time computer vision systems.