More Like this

1. Rhythmic pixel regions: multi-resolution visual sensing system towards high-precision visual computing at low power

   https://doi.org/10.1145/3445814.3446737  

   Kodukula, Venkatesh; Shearer, Alexander; Nguyen, Van; Lingutla, Srinivas; Liu, Yifei; LiKamWa, Robert (April 2021, Proceedings of the 26th ACM International Conference on Architectural Support for Programming Languages and Operating Systems)

   null (Ed.)

   High spatiotemporal resolution can offer high precision for vision applications, which is particularly useful to capture the nuances of visual features, such as for augmented reality. Unfortunately, capturing and processing high spatiotemporal visual frames generates energy-expensive memory traffic. On the other hand, low resolution frames can reduce pixel memory throughput, but reduce also the opportunities of high-precision visual sensing. However, our intuition is that not all parts of the scene need to be captured at a uniform resolution. Selectively and opportunistically reducing resolution for different regions of image frames can yield high-precision visual computing at energy-efficient memory data rates. To this end, we develop a visual sensing pipeline architecture that flexibly allows application developers to dynamically adapt the spatial resolution and update rate of different “rhythmic pixel regions” in the scene. We develop a system that ingests pixel streams from commercial image sensors with their standard raster-scan pixel read-out patterns, but only encodes relevant pixels prior to storing them in the memory. We also present streaming hardware to decode the stored rhythmic pixel region stream into traditional frame-based representations to feed into standard computer vision algorithms. We integrate our encoding and decoding hardware modules into existing video pipelines. On top of this, we develop runtime support allowing developers to flexibly specify the region labels. Evaluating our system on a Xilinx FPGA platform over three vision workloads shows 43 − 64% reduction in interface traffic and memory footprint, while providing controllable task accuracy. 

   more » « less
2. PeTra: A Sparsely Supervised Memory Model for People Tracking

   https://doi.org/10.18653/v1/2020.acl-main.481  

   Toshniwal, Shubham; Ettinger, Allyson; Gimpel, Kevin; Livescu, Karen (July 2020, Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics)

   We propose PeTra, a memory-augmented neural network designed to track entities in its memory slots. PeTra is trained using sparse annotation from the GAP pronoun resolution dataset and outperforms a prior memory model on the task while using a simpler architecture. We empirically compare key modeling choices, finding that we can simplify several aspects of the design of the memory module while retaining strong performance. To measure the people tracking capability of memory models, we (a) propose a new diagnostic evaluation based on counting the number of unique entities in text, and (b) conduct a small scale human evaluation to compare evidence of people tracking in the memory logs of PeTra relative to a previous approach. PeTra is highly effective in both evaluations, demonstrating its ability to track people in its memory despite being trained with limited annotation. 

   more » « less
3. HiRISE: High-Resolution Image Scaling for Edge ML via In-Sensor Compression and Selective ROI

   Reidy, B; Tabrizchi, S; Mohammadi, M; Angizi, S; Roohi, A; Zand, R (November 2024, ACM)

   With the rise of tiny IoT devices powered by machine learning (ML), many researchers have directed their focus toward compressing models to fit on tiny edge devices. Recent works have achieved remarkable success in compressing ML models for object detection and image classification on microcontrollers with small memory, e.g., 512kB SRAM. However, there remain many challenges prohibiting the deployment of ML systems that require high-resolution images. Due to fundamental limits in memory capacity for tiny IoT devices, it may be physically impossible to store large images without external hardware. To this end, we propose a high-resolution image scaling system for edge ML, called HiRISE, which is equipped with selective region-of-interest (ROI) capability leveraging analog in-sensor image scaling. Our methodology not only significantly reduces the peak memory requirements, but also achieves up to 17.7× reduction in data transfer and energy consumption. 

   more » « less
4. Multi-Resolution Real-Time Deep Pose-Space Deformation

   Zheng, Mianlun; Barbic, Jernej (November 2024, ACM Transactions on Graphics)

   We present a hard-real-time multi-resolution mesh shape deformation technique for skeleton-driven soft-body characters. Producing mesh deformations at multiple levels of detail is very important in many applications in computer graphics. Our work targets applications where the multi-resolution shapes must be generated at fast speeds (“hard-real-time”, e.g., a few milliseconds at most and preferably under 1 millisecond), as commonly needed in computer games, virtual reality and Metaverse applications. We assume that the character mesh is driven by a skeleton, and that high-quality character shapes are available in a set of training poses originating from a high-quality (slow) rig such as volumetric FEM simulation. Our method combines multi-resolution analysis, mesh partition of unity, and neural networks, to learn the pre-skinning shape deformations in an arbitrary character pose. Combined with linear blend skinning, this makes it possible to reconstruct the training shapes, as well as interpolate and extrapolate them. Crucially, we simultaneously achieve this at hard real-time rates and at multiple mesh resolution levels. Our technique makes it possible to trade deformation quality for memory and computation speed, to accommodate the strict requirements of modern real-time systems. Furthermore, we propose memory layout and code improvements to boost computation speeds. Previous methods for real-time approximations of quality shape deformations did not focus on hard real-time, or did not investigate the multi-resolution aspect of the problem. Compared to a "naive" approach of separately processing each hierarchical level of detail, our method offers a substantial memory reduction as well as computational speedups. It also makes it possible to construct the shape progressively level by level and interrupt the computation at any time, enabling graceful degradation of the deformation detail. We demonstrate our technique on several examples, including a stylized human character, human hands, and an inverse-kinematics-driven quadruped animal. 

   more » « less
5. HiRED: Attention-Guided Token Dropping for Efficient Inference of High-Resolution Vision-Language Models

   https://doi.org/10.1609/aaai.v39i2.32171  

   Arif, Kazi_Hasan Ibn; Yoon, JinYi; Nikolopoulos, Dimitrios S; Vandierendonck, Hans; John, Deepu; Ji, Bo (April 2025, Proceedings of the AAAI Conference on Artificial Intelligence)

   High-resolution Vision-Language Models (VLMs) are widely used in multimodal tasks to enhance accuracy by preserving detailed image information. However, these models often generate an excessive number of visual tokens due to the need to encode multiple partitions of a high-resolution image input. Processing such a large number of visual tokens poses significant computational challenges, particularly for resource-constrained commodity GPUs. To address this challenge, we propose High-Resolution Early Dropping (HiRED), a plug-and-play token-dropping method designed to operate within a fixed token budget. HiRED leverages the attention of CLS token in the vision transformer (ViT) to assess the visual content of the image partitions and allocate an optimal token budget for each partition accordingly. The most informative visual tokens from each partition within the allocated budget are then selected and passed to the subsequent Large Language Model (LLM). We showed that HiRED achieves superior accuracy and performance, compared to existing token-dropping methods. Empirically, HiRED-20% (i.e., a 20% token budget) on LLaVA-Next-7B achieves a 4.7x increase in token generation throughput, reduces response latency by 78%, and saves 14% of GPU memory for single inference on an NVIDIA TESLA P40 (24 GB). For larger batch sizes (e.g., 4), HiRED-20% prevents out-of-memory errors by cutting memory usage by 30%, while preserving throughput and latency benefits. 

   more » « less