VIDEO ANALYTICS CARD
DeePhi video analytics card solution (DP-2400-O12) with high-performance FPGA based Aristotle architecture, integrates with various of deep learning algorithms including vehicle/pedestrian detection, model and plate number recognition and attributes recognition. It is able to achieve real-time structural information analysis of motorized/non-motorized vehicles and pedestrians from multi-channels video stream.
Video Structuralization
Vehicle/non-motor vehicle/pedestrian and all moving targets’ attribute analytics
Vehicle Type Recognition
SUV, coach, minibus, truck, sedan, van, etc.
Vehicle Attribute Recognition
Color, plate, direction
Pedestrian Attribute Recognition
Gender, direction, clothes color, etc.
H.265/H.264 Decode
4x4K*2K(3840*2160)@30fps
H.264 Encode
2x1080p@30fps
JPEG Encode
4x4K*2K@2fps
H.265/H.264 Decode
32x720p@30fps
H.264 Encode
4x720p@30fps
JPEG Encode
32x720p@2fps
H.265/H.264 Decode
16x1080p@30fps
H.264 Encode
2x1080p@30fps
JPEG Encode
16x1080p@2fps
H.265/H.264 Decode
64xD1@30fps
H.264 Encode
8xD1@30fps
JPEG Encode
64xD1@2fps
H.265/H.264 Decode
9x1080p@30fps
H.264 Encode
4Kx2K@30fps
JPEG Encode
9x1080p@2fps
JPEG Encode
9x720p@30fps
Up to 12-channel real-time video analytics with 1080p resolution.
CHANNELS
Video Stream
12CH@1080p
SPECIFICATION
Vehicle Attribute
Model
SUV, coach, minibus, truck, sedan, van;
Vehicle color
White, grey, yellow, magenta, red, purple, green, blue, brown, black, other
Plate color
Blue, yellow, white, black
Plate number recognition
Pedestrian Attribute
Clothes color
White, grey, yellow, magenta, red, purple, green, blue, brown, black, other
Hat, gender, package
Vehicle Detection
Real-time vehicle detection and bounding box regression from video stream
Pedestrian Detection
Real-teme pedestrian detetion and bounding box regression from video stream
Non-motor Vehicle Detection
Real-time non-motor vehicles detection and bounding box regression from video stream
CONNECTIVITY
API & SDK
C/C++ API and SDK, integratable to third-party platform
HW Interface
PCIe/RJ45
GENERAL
Image Format
JPEG/JPG/BMP/PNG
Encode Format
H.264/H.265
Dimension
185mm*112mm*35mm
Power Consumption
20~30w
Temperature
0℃~70℃
Humidity
10%~90% (non-condensing)