Image Signal Processors (ISPs) convert raw sensor signals into digital images, which significantly influence the image quality and the performance of downstream computer vision tasks. Designing ISP pipeline and tuning ISP parameters are two key steps for building an imaging and vision system. To find optimal ISP configurations, recent works use deep neural networks as a proxy to search for ISP parameters or ISP pipelines. However, these methods are primarily designed to maximize the image quality, which are sub-optimal in the performance of high-level computer vision tasks such as detection, recognition, and tracking. Moreover, after training, the learned ISP pipelines are mostly fixed at the inference time, whose performance degrades in dynamic scenes. To jointly optimize ISP structures and parameters, we propose AdaptiveISP, a task-driven and scene-adaptive ISP. One key observation is that for the majority of input images, only a few processing modules are needed to improve the performance of downstream recognition tasks, and only a few inputs require more processing. Based on this, AdaptiveISP utilizes deep reinforcement learning to automatically generate an optimal ISP pipeline and the associated ISP parameters to maximize the detection performance.

Image Signal Processors (ISPs) convert raw sensor signals into digital images, which significantly influence the image quality and the performance of downstream computer vision tasks. Designing ISP pipeline and tuning ISP parameters are two key steps for building an imaging and vision system. To find optimal ISP configurations, recent works use deep neural networks as a proxy to search for ISP parameters or ISP pipelines. However, these methods are primarily designed to maximize the image quality, which are sub-optimal in the performance of high-level computer vision tasks such as detection, recognition, and tracking. Moreover, after training, the learned ISP pipelines are mostly fixed at the inference time, whose performance degrades in dynamic scenes. To jointly optimize ISP structures and parameters, we propose AdaptiveISP, a task-driven and scene-adaptive ISP.

RAW image datasets are more suitable than the standard RGB image datasets for the ill-posed inverse problems in low-level vision, but not common in the literature. There are also a few studies to focus on mapping sRGB images to RAW format. Mapping from sRGB to RAW format could be a relevant domain for reverse style transferring since the task is an ill-posed reversing problem. In this study, we seek an answer to the question: Can the ISP operations be modeled as the style factor in an end-to-end learning pipeline? To investigate this idea, we propose a novel architecture, namely RST-ISP-Net, for learning to reverse the ISP operations with the help of adaptive feature normalization. We formulate this problem as a reverse style transferring and mostly follow the practice used in the prior work. We have participated in the AIM Reversed ISP challenge with our proposed architecture. Results indicate that the idea of modeling disruptive or modifying factors as style is still valid, but further improvements are required to be competitive in such a challenge.

Arm has introduced a design for an automotive image signal processor to enhance driver assistance and automation technologies. The Arm Mali-C78AE image signal processor (ISP) is part of Arm's AE line of safety-capable intellectual property suitable for advanced drivers assistance systems (ADAS) and human vision applications. It's the first product announcement since Nvidia called off the $80 billion acquisition of Arm last week. The first licensee for the tech is Intel's Mobileye, which is licenses the Mali-C78AE and the next-generation EyeQ technology. ADAS tech has grown from a premium vehicle feature to a capability consumers now expect as standard in new vehicles, as the systems have helped with driver safety.

Every December for the last few years, Qualcomm has held an annual event in Hawaii to announce its latest flagship mobile chipset. This year was no different with the company taking the opportunity to unveil the Snapdragon 8 Gen 1. That's right, for the second year in a row, Qualcomm is moving away from the sequential numbering scheme that has defined its processors for years. Just as the Snapdragon 865 gave way to the 888, the company will now replace the 888 with the Gen 1. The company says it's capable of theoretical download speeds of 10Gbps. That's one of those specs that's impressive on paper, but won't mean much out in the real world since some of the fastest 5G networks can't deliver speeds greater than 4Gbps in ideal conditions.

NoC interconnect IP to be dataflow backbone for image signal processors providing enhanced sensitivity, high-resolution HD imaging through low current, low power in a single-chip solution for the security/surveillance market. Eyenix's imaging solution provides a step-function advance over their previous product, replacing a 3rd-party artificial intelligence (AI) function with a superior capability developed in-house for super-resolution imaging. This is provided in a tightly integrated system design, including functions for image stabilization for mobile usage and image dewarping for wide-angle camera correction. The first application is destined for surveillance camera applications. Eyenix chose Arteris IP on-chip interconnect technology as a part of Eyenix's proprietary image processing chip because it enables Eyenix to design and integrate a complete and superior Eyenix imaging solution without dependency on external IP blocks for the AI function.

VANCOUVER, British Columbia, July 20, 2020 (GLOBE NEWSWIRE) – NexOptic Technology Corp. ("NexOptic" or the "Company") (TSX VENTURE: NXO) (OTCQB: NXOPF) (FSE: E3O1), an optics and Artificial Intelligence ("AI") innovator, announced that it has created revolutionary AI that transforms Image Signal Processors ("ISP") technology. Engineered into NexOptic's All Light Intelligent Imaging Solutions ("ALIIS "), the neural ISP technology is immediately available to OEM customers through Nexoptic's alliances with leading semiconductor companies. Image signal processors are a traditional technology that manipulate images from raw data into the precise and coherent imagery we are accustomed to seeing. They are increasingly also being used for new application paths in robotics, smart cities, industrial automation, automotive, healthcare and beyond. NexOptic's game-changing neural ISP seamlessly delivers this technology into the world of AI for the first time.

Among the challenges facing autonomous vehicle developers is the need to gather and process vast amounts of data gathered by cameras quickly and efficiently. Last month, BlinkAI Technologies Inc. announced its RoadSight product, which is designed to improve camera performance in low-light conditions. While some autonomous vehicle makers are using multiple cameras rather than more expensive lidar technology, that approach has raised safety concerns. The National Traffic Safety Board recently found that a contributing factor to the fatal Uber crash in 2018 was that the automated driver system did not recognize a jaywalking pedestrian in a low-light setting. BlinkAI spun out from the MIT-Harvard Martinos Center of biomedical imaging and emerged from "stealth mode" over the past few months.

Expensive flagship phones won't be the only way for you to play with advanced features like AR Emoji, Animoji and Face ID much longer. Qualcomm is making it easier for companies to create mid-range smartphones that pack those functions by launching a new mobile processor. The Snapdragon 710 will come with a multi-core AI Engine and support neural network processing, as well as image signal processors and graphics units that are typically found in higher-end chipsets. The 710 is the first of the 700-series, which was announced at MWC this year, and will sit above options like the 600- and 400-ranges but below top-tier chips like the Snapdragon 845. The Snapdragon 710 is a 10nm chipset that features a multi-core AR engine for on-device neural networking processing, as well as a Spectra 250 image signal processor that enables things like multi-frame noise reduction and AI camera features like video style transfer and active depth sensing for artificial bokeh.

Artificial intelligence is permeating everybody's lives through the face recognition, voice recognition, image analysis and natural language processing capabilities built into their smartphones and consumer appliances. Over the next several years, most new consumer devices will run AI natively, locally and, to an increasing extent, autonomously. But there's a problem: Traditional processors in most mobile devices aren't optimized for AI, which tends to consume a lot of processing, memory, data and battery on these resource-constrained devices. As a result, AI has tended to execute slowly on mobile and "internet of things" endpoints, while draining their batteries rapidly, consuming inordinate wireless bandwidth and exposing sensitive local information as data makes roundtrips in the cloud. That's why mass-market mobile and IoT edge devices are increasingly coming equipped with systems-on-a-chip that are optimized for local AI processing.