Tipo de dados
Picture or Video
Supported Algorithm Type
Large Model: Single detection, combined algorithm, dual detection;
Small Model: Classification, OCR (Optical Character Recognition).
Picture Resolution Requirement
32×32 to 4096×4096 Picture format: JPEG(JPG)、TIF、BMP、PNG、gif (static);Picture size: no more than 10 MB.
Exigência de taxa de quadros de vídeo
H.264 video stream、H.265 video stream (HIKVISION),Resolution: 2 MP-8 MP(4096 × 2160);
Frame rate: 12/s to 30/s;
Stream packaging format supported by video importing: ps、ts、mp4、avi、mbf、asf，h.264、h.265(HIKVISION)、svac、hik es、Dahua es、Tiandi platform es;
Video size: no more than 100GB in total, and no more than 10GB per video.
Picture Analysis Performance
Performance specification (2 MP resolution with less than 10 targets in the scene):Object detection, single tag classification, multiple tags classification, multiple attributes classification, mix (1 for detection, 3 for classification) and 256 pictures per second.
Desempenho de análise de vídeo
Performance specification (2 MP to 8 MP (4096*2160)：resolution-insensitive; tracking 64 targets in scene at the same time and outputting one target per second):Object detection, single tag classification, multiple tags classification, multiple attributes classification, mix (1 for detection, 3 for classification) and 16 channels video analysis.
Picture rotation analysis
Performance specification (2 MP resolution with less than 10 targets in the scene):
Object detection, single tag classification, multiple tags classification, multiple attributes classification, mix (1 for detection, 3 for classification).
Support up to 16 tasks of scheduled capture analysis (each capture task can link to 16 channels rotation analysis, capture interval can be 1 to 3600 seconds).
