FPGA AI Suite

Altera FPGA AI suite provides a complete software solution to quickly convert and optimize trained AI model to custom AI inference accelerator and deploy on Altera FPGAs enabling seamless collaboration between data scientists and FPGA engineers with easy-to-use tools.

Altera FPGA AI suite provides:

• Support popular AI frameworks: The front-end of FPGA AI suite is OpenVINO Toolkit. The Toolkit seamlessly converts trained (inference) models from most standard frameworks like Pytorch and TensorFlow to intermediate representation that FPGA AI suite uses.
• Hardware-less early model validation: Data scientists can do quick check on trained model’s performance, model accuracy, power, and area in FPGA without hardware. FPGA AI suite’s model optimizer auto optimizes the IP parameters for target area or performance.
• Access to free pre-trained AI models: Jumpstart solution development by picking the best suited model. Support for most of the deep learning models on Open Model Zoo for OpenVINO toolkit.
• Ease of use: Push-button AI inference IP generation.
• Easy integration of AI inference IP: Add AI inference IP to FPGA design with best-in-class standard Quartus® prime design software FPGA flows.
• Lower TCO (total cost of ownership) with easy system integration: Integrate AI IP with other system level components to achieve smaller footprint, lower power, low latency, and system costs.

Applications

• Computer vision: image classification, object detection, semantic segmentation, pose estimation, anomaly detection, predictive maintenance
• Medical imaging and diagnosis: endoscopy, ultrasounds, CT/PET/MR/X-rays
• Industrial: predictive maintenance, quality inspection and assurance
• Telecom: adaptive beamforming, network security, network optimization, fronthaul traffic reduction, packet inspection
• Military: target detection (e.g. missile), RF sensing for surveillance, unmanned aerial vehicles
• Broadcast- video conferencing