Based on AMD Zynq UltraScale+ MPSoC + NVIDIA Jetson Orin NX heterogeneous design, specifically designed for industrial computer vision needs. The platform adopts YOLOX object recognition algorithm, providing extremely high detection accuracy and speed, with mAP up to 0.9 and recognition speed up to 50FPS.
· High precision recognition: YOLOX object recognition algorithm provides high-precision recognition in industrial environments, ensuring mAP of 0.9
· Multi object detection: supports multi object detection, can accurately identify and classify various electronic devices, and improve detection efficiency
· High speed processing: With a processing capacity of 50 frames per second, it ensures the continuity and efficiency of industrial production lines
· Real time brightness adjustment
· Digital lossless zoom
· Automatic balancing and calibration
· High resolution photography
· Image Embedding
· Color correction and 3D parameter calibration
· Left and right eye frame switching and cross contrast adjustment
· Supports H.264 and H.265 encoding and decoding: providing high-quality video compression and storage solutions
· Real time transmission: supports real-time video data transmission to meet the real-time monitoring and analysis needs of industrial sites
· External storage device compatibility: supports multiple external storage devices to meet the needs of large capacity data storage
· FPGA and GPU working together: improving data processing efficiency and ensuring system real-time performance
· Parallel processing: supports multitasking parallel processing, greatly improving computational efficiency
· Powerful computing power: The GPU computing power provided by NVIDIA Jetson Orin NX is suitable for high-density data processing
· Close up mode: suitable for detailed inspection of small electronic devices
· Ultra close range mode: suitable for fine detection of extremely small components
ALINX is designed based on the heterogeneous architecture of AMD Zynq UltraScale+MPSoC and NVIDIA Jetson Orin NX, using YOLOX object recognition algorithm to provide extremely high detection accuracy and speed, with mAP up to 0.9 and recognition speed up to 50FPS. It is applied in semiconductor testing, electronic manufacturing, photovoltaic energy, lithium batteries, industrial automation, industrial testing, assembly and other fields. By using NVIDIA Jetson Orin NX to process and analyze image information, multiple functions such as recognition, detection, tracking, and measurement of target objects can be achieved.
Quickly and accurately detect various electrical components, circuit defects, and other quality issues on PCB boards to improve production efficiency
Industrial robot manufacturing and assembly, electronic product placement, etc. on the production line meet the requirements of high speed and high precision
Real time detection and troubleshooting in automated production lines to reduce downtime
Tasks such as automatic recognition, sorting, transmission, environmental perception, gesture recognition, etc. for electronic components, parts, drugs, etc
Quality inspection of each stage of the production process through machine vision to ensure the quality of each product is qualified
Assist industrial robots in navigating and operating in complex environments, improving operational accuracy and efficiency
Based on the heterogeneous design of AMD Zynq UltraScale+MPSoC+NVIDIA Jetson Orin NX, it meets the requirements of high-speed data transmission, AI edge computing, professional high-definition 8K video image acquisition and processing, rich 12G-SDI, DP, HDMI video interfaces, 100G optical fiber interfaces, and is suitable for industrial and manufacturing, medical endoscopy and surgical robots, robot development, smart city, security and other fields of solutions
FPGA Chip
GPU
AI computility
8-channel bidirectional interface
HDMI 2.1 Input/Output
2-channel fiber optic interface
Logic Cells
PS DDR4 | PL DDR4
FPGA
Device：XCZU19EG-2FFVC1760I，packaged： FFVC1760
Logic Cells：1,143K，Flip-Flops：1,045K，LUTs：523K，Block RAM：34.6Mb，DSP Slices：1968，GTH：16.3Gb/s x44
CPU
Application processing unit：4x ARM Cortex-A53, 1.3GHz；
Real time processing unit：Dual core Cortex™-R5, 533MHz
GPU
Mali™-400 MP2，Frequency up to 667MHz
Cache | Storage
PS 8GB DDR4，PL 16GB DDR4 | 32GB eMMC FLASH，128MB QSPI FLASH
12G SDI Interface
8-channel bidirectional 12G SDI interface, supporting SD/HD/3G/6G/12G SDI rates, supporting up to 4K/60 frames of video input or output
100G Fiber optic interface
2 QSFP100G fiber optic interfaces, with a total data rate of 100Gbps per channel
Other interfaces
1 expansion port for FMC HPC, connecting various FMC modules externally; Gigabit Ethernet, JTAG, Uart and other interfaces
model
Jetson Orin NX 8GB (Optional selection)
Jetson Orin NX 16GB (Optional selection)
computility
70TOPS
100TOPS
CPU
6 cores NVIDIA Arm Cortex A78AE V8.2 64 bit CPU，2GHz
8 cores NVIDIA Arm Cortex A78AE V8.2 64 bit CPU，2GHz
GPU
1024 cores NVIDIA Ampere GPU with 32 Tensor Cores
1024 cores NVIDIA Ampere GPU with 32 Tensor Cores
Video Memory
8GB 128bit LPDDR5
16GB 128bit LPDDR5
video coding
1x 4K60 (H.265); 3x 4K30 (H.265 6x 1080p60 (H.265); 12x 1080p30 (H.265)
Video decoding
1x 8K30 (H.265); 2x 4K60 (H.265); 4x 4K30 (H.265); 9x 1080p60 (H.265); 18x 1080p30 (H.265)
DP display interface
1-channel Display Port output interface, used for displaying video images, supports up to 4K@30Hz or 1080P@60Hz Output
Other interfaces
NVMe M.2 SSD hard drive (self purchased) M.2 WIFI/BT、USB3.0、 Gigabit Ethernet and other interfaces
Links： AMD Pangomicro AUMO Autonomous Driving
Alinx Electronic Limited 沪ICP备13046728号
