{"product_id":"pva-accelerated-isp","title":"PVA Accelerated ISP","description":"\u003ch2\u003eSoftware-defined Image Signal Processor (ISP) accelerated by NVIDIA's Programmable Vision Accelerator (PVA)\u003cspan class=\"m_3377163651073292963qb m_3377163651073292963wb\"\u003e\u003c\/span\u003e\n\u003c\/h2\u003e\n\u003cp\u003eRidgeRun's PVA ISP is designed for teams that need greater control over image processing than NVIDIA Argus exposes, without consuming GPU resources with a CUDA-based ISP. It preserves GPU capacity for AI, computer vision, and other compute-intensive workloads.\u003cspan aria-hidden=\"true\" class=\"PDq2pG_selectionAnchor\"\u003e\u003c\/span\u003e\u003c\/p\u003e\n\u003cp\u003eThe pipeline includes essential ISP stages such as black level correction, demosaicing, anti-aliasing, auto white balance, auto exposure, gamma correction, and tone mapping. Executing these operations on dedicated PVA hardware enables efficient, low-power RAW Bayer processing for real-time embedded camera applications.\u003c\/p\u003e\n\u003cp\u003eThe implementation is designed to provide high throughput with low latency and power consumption:\u003c\/p\u003e\n\u003cul\u003e\n\u003cli\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003eProcesses up to \u003cstrong data-start=\"701\" data-end=\"742\"\u003etwo simultaneous 4K streams at 30 fps\u003c\/strong\u003e or \u003cstrong data-start=\"746\" data-end=\"779\"\u003eeight 1080p streams at 30 fps\u003c\/strong\u003e\u003c\/span\u003e\u003c\/li\u003e\n\u003cli\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003eProcesses each 1080p frame in approximately \u003cstrong data-start=\"826\" data-end=\"834\"\u003e8 ms\u003c\/strong\u003e\u003c\/span\u003e\u003c\/li\u003e\n\u003cli data-section-id=\"1xk983n\" data-start=\"835\" data-end=\"891\" style=\"color: rgb(0, 0, 0);\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003eConsumes approximately \u003cstrong data-start=\"860\" data-end=\"891\"\u003e0.75 W for one 1080p stream\u003c\/strong\u003e\u003c\/span\u003e\u003c\/li\u003e\n\u003cli data-section-id=\"118t4v8\" data-start=\"892\" data-end=\"964\" style=\"color: rgb(0, 0, 0);\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003eUses \u003cstrong data-start=\"899\" data-end=\"917\"\u003e0% GPU compute\u003c\/strong\u003e, keeping the GPU available for other workloads\u003c\/span\u003e\u003c\/li\u003e\n\u003cli data-section-id=\"2xspv0\" data-start=\"965\" data-end=\"1046\" style=\"color: rgb(0, 0, 0);\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003eSupports \u003cstrong data-start=\"976\" data-end=\"995\"\u003ezero-copy input\u003c\/strong\u003e to reduce memory transfers and processing overhead\u003c\/span\u003e\u003c\/li\u003e\n\u003cli data-section-id=\"1ha4rbe\" data-start=\"1047\" data-end=\"1113\" style=\"color: rgb(0, 0, 0);\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003eDelivers latency comparable to NVIDIA's Argus-based ISP pipeline\u003c\/span\u003e\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eThese results make PVA ISP suitable for high-resolution and multi-camera systems where processing performance, power efficiency, and GPU availability are critical. Actual performance and power consumption may vary depending on the Jetson platform, enabled ISP stages, and pipeline configuration.\u003c\/p\u003e\n\u003cp\u003eBeyond performance, developers gain visibility and control over the complete processing workflow. Individual stages can be configured, tuned, replaced, reordered, or extended, with access to intermediate results and support for custom algorithms optimized for image quality or machine perception.\u003c\/p\u003e\n\u003cp\u003eThe framework transforms RAW Bayer sensor data into formats suitable for display, video encoding, streaming, and computer vision. It can be integrated through a GStreamer element for rapid prototyping and pipeline-based deployment, or through its C++ API for deeper customization and direct control.\u003c\/p\u003e\n\u003cp\u003eUnlike the fixed-function ISP pipeline exposed through NVIDIA Argus, RidgeRun's PVA ISP enables application-specific imaging workflows while retaining the performance and power-efficiency benefits of dedicated hardware acceleration.\u003c\/p\u003e\n\u003chr\u003e\n\u003ch3\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003eKey Features:\u003c\/span\u003e\u003c\/h3\u003e\n\u003cul\u003e\n\u003cli style=\"color: rgb(0, 0, 0);\"\u003e\n\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cstrong data-end=\"83\" data-start=\"53\"\u003e\u003cstrong data-is-only-node=\"\" data-end=\"28\" data-start=\"2\"\u003eComplete ISP pipeline\u003c\/strong\u003e\u003c\/strong\u003e\u003c\/span\u003e\u003cbr data-end=\"86\" data-start=\"83\"\u003eIncludes black level correction, digital gain, demosaicing, anti-aliasing, auto white balance, auto exposure, gamma correction, and tone mapping.\u003cbr\u003e\n\u003c\/li\u003e\n\u003cli style=\"color: rgb(0, 0, 0);\"\u003e\n\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cstrong data-end=\"83\" data-start=\"53\"\u003ePVA-Accelerated Processing\u003c\/strong\u003e\u003c\/span\u003e\u003cbr data-end=\"86\" data-start=\"83\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003eExecutes ISP algorithms on the NVIDIA Programmable Vision Accelerator, freeing CPU and GPU resources.\u003cspan class=\"PDq2pG_selectionAnchor\" aria-hidden=\"true\"\u003e\u003c\/span\u003e\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"color: rgb(0, 0, 0);\"\u003e\n\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cstrong data-end=\"226\" data-start=\"193\"\u003eSoftware-Defined ISP Pipeline\u003c\/strong\u003e\u003c\/span\u003e\u003cbr data-end=\"229\" data-start=\"226\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003eProvides a configurable alternative to fixed-function and black-box ISP implementations.\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"color: rgb(0, 0, 0);\"\u003e\n\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cstrong data-end=\"528\" data-start=\"501\"\u003eCustomizable Algorithms\u003c\/strong\u003e\u003c\/span\u003e\u003cbr data-end=\"531\" data-start=\"528\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003eAllows ISP stages to be tuned, replaced, reordered, or extended for application-specific image-quality requirements.\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"color: rgb(0, 0, 0);\"\u003e\n\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cstrong data-end=\"814\" data-start=\"778\"\u003eLinear and Companded RAW Support\u003c\/strong\u003e\u003c\/span\u003e\u003cbr data-end=\"817\" data-start=\"814\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003eSupports linear Bayer data and companded HDR RAW input through configurable decompanding.\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"color: rgb(0, 0, 0);\"\u003e\n\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cstrong data-end=\"939\" data-start=\"912\"\u003eMultiple Output Formats\u003c\/strong\u003e\u003c\/span\u003e\u003cbr data-end=\"942\" data-start=\"939\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003eProduces standard formats suitable for downstream processing, including RGB888 and NV12.\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"color: rgb(0, 0, 0);\"\u003e\n\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cstrong data-end=\"1059\" data-start=\"1033\"\u003eSensor-Agnostic Design\u003c\/strong\u003e\u003c\/span\u003e\u003cbr data-end=\"1062\" data-start=\"1059\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003eWorks with Bayer RAW sensors independently of the camera manufacturer.\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"color: rgb(0, 0, 0);\"\u003e\n\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cstrong data-end=\"1174\" data-start=\"1138\"\u003eRuntime Configuration and Tuning\u003c\/strong\u003e\u003c\/span\u003e\u003cbr data-end=\"1177\" data-start=\"1174\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003eISP parameters can be adjusted without redesigning a fixed hardware pipeline.\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"color: rgb(0, 0, 0);\"\u003e\n\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cstrong data-end=\"1290\" data-start=\"1260\"\u003eLow-Power Image Processing\u003c\/strong\u003e\u003c\/span\u003e\u003cbr data-end=\"1293\" data-start=\"1290\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003eUses dedicated vision hardware to reduce reliance on power-intensive GPU processing.\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"color: rgb(0, 0, 0);\"\u003e\n\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cstrong data-end=\"1417\" data-start=\"1383\"\u003eReal-Time Embedded Integration\u003c\/strong\u003e\u003c\/span\u003e\u003cbr data-end=\"1420\" data-start=\"1417\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003eDesigned for low-latency camera pipelines on supported NVIDIA Jetson platforms equipped with PVA.\u003c\/span\u003e\n\u003c\/li\u003e\n\u003cli style=\"color: rgb(0, 0, 0);\"\u003e\n\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cstrong data-end=\"1551\" data-start=\"1523\"\u003eSource-Code Availability\u003c\/strong\u003e\u003c\/span\u003e\u003cbr data-end=\"1554\" data-start=\"1551\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003eEnables customers to inspect, tune, replace, and add algorithms when purchased with source-code access.\u003c\/span\u003e\n\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003chr\u003e\n\u003ch3 dir=\"ltr\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003eExample Use Cases\u003c\/span\u003e\u003c\/h3\u003e\n\u003cul\u003e\n\u003cli data-end=\"163\" data-start=\"0\" data-section-id=\"x6z43y\" style=\"color: rgb(0, 0, 0);\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cstrong data-is-only-node=\"\" data-end=\"40\" data-start=\"2\"\u003eMulti-camera perception platforms:\u003c\/strong\u003e Offloads ISP processing from the GPU, preserving resources for perception and AI workloads across multiple camera streams.\u003cspan class=\"PDq2pG_selectionAnchor\" aria-hidden=\"true\"\u003e\u003c\/span\u003e\u003c\/span\u003e\u003c\/li\u003e\n\u003cli data-end=\"462\" data-start=\"299\" data-section-id=\"1cu0sg5\" style=\"color: rgb(0, 0, 0);\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cstrong data-is-only-node=\"\" data-end=\"356\" data-start=\"301\"\u003eCustom processing pipelines for machine perception:\u003c\/strong\u003e Enables algorithms to be tuned for feature preservation and model accuracy rather than visual appearance.\u003c\/span\u003e\u003c\/li\u003e\n\u003cli data-end=\"663\" data-start=\"464\" data-section-id=\"i8vieb\" style=\"color: rgb(0, 0, 0);\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cstrong data-is-only-node=\"\" data-end=\"538\" data-start=\"466\"\u003eFully controlled imaging pipelines for safety-critical applications:\u003c\/strong\u003e Provides visibility and control over each processing stage, supporting deterministic behavior, validation, and traceability.\u003c\/span\u003e\u003c\/li\u003e\n\u003cli data-end=\"813\" data-start=\"665\" data-section-id=\"j1ekdb\" style=\"color: rgb(0, 0, 0);\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cstrong data-is-only-node=\"\" data-end=\"700\" data-start=\"667\"\u003eOffline RAW image processing:\u003c\/strong\u003e Processes previously captured Bayer RAW data for tuning, testing, dataset preparation, and algorithm validation.\u003c\/span\u003e\u003c\/li\u003e\n\u003cli data-end=\"944\" data-start=\"815\" data-section-id=\"tmh0e7\" style=\"color: rgb(0, 0, 0);\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cstrong data-is-only-node=\"\" data-end=\"854\" data-start=\"817\"\u003eLow-power edge AI camera systems:\u003c\/strong\u003e Uses the dedicated PVA hardware to reduce GPU usage and improve overall power efficiency.\u003c\/span\u003e\u003c\/li\u003e\n\u003cli data-is-last-node=\"\" data-end=\"1098\" data-start=\"946\" data-section-id=\"rxoish\" style=\"color: rgb(0, 0, 0);\"\u003e\n\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cstrong data-is-only-node=\"\" data-end=\"983\" data-start=\"948\"\u003eHigh-resolution camera systems:\u003c\/strong\u003e Accelerates computationally intensive ISP operations to support high-resolution image processing with low latency.\u003c\/span\u003e\u003cbr\u003e\n\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003chr\u003e\n\u003ch3 dir=\"ltr\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003eSupported Platforms\u003c\/span\u003e\u003c\/h3\u003e\n\u003cul\u003e\n\u003cli style=\"color: rgb(0, 0, 0);\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cspan style=\"font-size: 0.875rem;\"\u003eNVIDIA Jetson \u003cstrong\u003eThor.\u003c\/strong\u003e\u003c\/span\u003e\u003c\/span\u003e\u003c\/li\u003e\n\u003cli role=\"presentation\" dir=\"ltr\" style=\"color: rgb(0, 0, 0);\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003eNVIDIA Jetson \u003cstrong\u003e\u003cspan style=\"font-size: 0.875rem;\"\u003eAGX Orin.\u003c\/span\u003e\u003c\/strong\u003e\u003c\/span\u003e\u003c\/li\u003e\n\u003cli role=\"presentation\" dir=\"ltr\" style=\"color: rgb(0, 0, 0);\"\u003e\u003cspan style=\"font-size: 0.875rem; color: rgb(0, 0, 0);\"\u003eNVIDIA Jetson \u003cstrong\u003eOrin NX.\u003c\/strong\u003e\u003c\/span\u003e\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cspan style=\"font-size: 0.875rem;\"\u003ePlatform compatibility is determined by the availability of supported NVIDIA PVA hardware.\u003c\/span\u003e\u003cspan style=\"font-size: 0.875rem;\"\u003e\u003c\/span\u003e\u003c\/span\u003e\u003c\/p\u003e\n\u003chr\u003e\n\u003ch3 dir=\"ltr\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003eWhat’s Included\u003c\/span\u003e\u003c\/h3\u003e\n\u003cul\u003e\n\u003cli data-end=\"159\" data-start=\"21\" data-section-id=\"llhv7g\" style=\"color: rgb(0, 0, 0);\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cstrong data-end=\"39\" data-start=\"23\"\u003eC++ library:\u003c\/strong\u003e Core PVA-accelerated ISP library with APIs for configuring, integrating, and controlling the image-processing pipeline.\u003cspan class=\"PDq2pG_selectionAnchor\" aria-hidden=\"true\"\u003e\u003c\/span\u003e\u003c\/span\u003e\u003c\/li\u003e\n\u003cli data-end=\"308\" data-start=\"161\" data-section-id=\"1462v1x\" style=\"color: rgb(0, 0, 0);\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cstrong data-end=\"188\" data-start=\"163\"\u003eExample applications:\u003c\/strong\u003e Ready-to-use examples demonstrating pipeline configuration, image processing, and integration into custom applications.\u003c\/span\u003e\u003c\/li\u003e\n\u003cli data-end=\"445\" data-start=\"310\" data-section-id=\"1club4c\" style=\"color: rgb(0, 0, 0);\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cstrong data-end=\"334\" data-start=\"312\"\u003eGStreamer element:\u003c\/strong\u003e A GStreamer plugin that simplifies integration and deployment within existing multimedia and camera pipelines.\u003c\/span\u003e\u003c\/li\u003e\n\u003cli data-is-last-node=\"\" data-end=\"606\" data-start=\"447\" data-section-id=\"16g2jij\" style=\"color: rgb(0, 0, 0);\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cstrong data-end=\"484\" data-start=\"449\"\u003eDebugging and calibration tool:\u003c\/strong\u003e A visual tool for inspecting intermediate results, adjusting ISP parameters, and tuning image quality during development.\u003c\/span\u003e\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003chr\u003e\n\u003ch3 dir=\"ltr\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003eWhy RidgeRun PVA ISP?\u003c\/span\u003e\u003c\/h3\u003e\n\u003cul\u003e\n\u003cli data-section-id=\"3e2peo\" data-start=\"27\" data-end=\"187\" style=\"color: rgb(0, 0, 0);\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cstrong data-start=\"29\" data-end=\"62\"\u003eBacked by RidgeRun expertise:\u003c\/strong\u003e Built by a team with extensive experience in GStreamer, embedded Linux, camera systems, and hardware-accelerated multimedia.\u003cspan aria-hidden=\"true\" class=\"PDq2pG_selectionAnchor\"\u003e\u003c\/span\u003e\u003c\/span\u003e\u003c\/li\u003e\n\u003cli data-section-id=\"152zefx\" data-start=\"189\" data-end=\"308\" style=\"color: rgb(0, 0, 0);\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cstrong data-start=\"191\" data-end=\"214\"\u003eFree GPU resources:\u003c\/strong\u003e Offload image processing to the PVA, preserving GPU capacity for AI and perception workloads.\u003c\/span\u003e\u003c\/li\u003e\n\u003cli data-section-id=\"1rbqnjw\" data-start=\"310\" data-end=\"449\" style=\"color: rgb(0, 0, 0);\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cstrong data-start=\"312\" data-end=\"343\"\u003eGain full pipeline control:\u003c\/strong\u003e Inspect, configure, tune, replace, and extend ISP stages without depending on a black-box implementation.\u003c\/span\u003e\u003c\/li\u003e\n\u003cli data-section-id=\"ryle5a\" data-start=\"451\" data-end=\"569\" style=\"color: rgb(0, 0, 0);\"\u003e\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cstrong data-start=\"453\" data-end=\"482\"\u003eReduce power consumption:\u003c\/strong\u003e Use dedicated PVA hardware for efficient processing in power-constrained edge systems.\u003c\/span\u003e\u003c\/li\u003e\n\u003cli data-section-id=\"ntozo1\" data-start=\"571\" data-end=\"709\" data-is-last-node=\"\" style=\"color: rgb(0, 0, 0);\"\u003e\n\u003cspan style=\"color: rgb(0, 0, 0);\"\u003e\u003cstrong data-start=\"573\" data-end=\"618\"\u003eBuild application-specific ISP workflows:\u003c\/strong\u003e Adapt the pipeline to your sensor, image-quality, HDR, or machine-perception requirements.\u003c\/span\u003e\u003cbr\u003e\n\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp dir=\"ltr\"\u003e\u003cspan\u003eLearn more in our \u003ca title=\"PVA ISP for NVIDIA Jetson\" href=\"https:\/\/developer.ridgerun.com\/wiki\/index.php\/PVA_ISP_for_NVIDIA_Jetson\" target=\"_blank\"\u003edeveloper's wiki.\u003c\/a\u003e\u003c\/span\u003e\u003c\/p\u003e","brand":"RidgeRun","offers":[{"title":"Binary blob","offer_id":48031407702203,"sku":null,"price":4999.0,"currency_code":"USD","in_stock":true},{"title":"Binary blob with source code","offer_id":48031407734971,"sku":null,"price":6999.0,"currency_code":"USD","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0051\/2239\/2135\/files\/Disenosintitulo_58.png?v=1782487160","url":"https:\/\/shop.ridgerun.com\/products\/pva-accelerated-isp","provider":"RidgeRun","version":"1.0","type":"link"}