File Type | Limit |
Image | PNG, JPEG, JPG, BMP, WEBP, GIF, HEIF, and HEIC formats are supported. Image files below 32 MB are supported for moderation. For images above 5MB, use the large-image-detect parameter when calling requests. Image resolution should be no less than 20 x 20 and no more than 40,000 x 40,000 (synchronous moderation no more than 10,000 x 10,000). It is recommended that the resolution be greater than 256 x 256, otherwise the recognition effect may be affected. |
Video | MP4, AVI, MKV, WMV, RMVB, FLV, and M3U8 formats are supported. The video size cannot exceed 5 GB, and the number of captured frames cannot exceed 10,000. |
Audio | Audio format: Currently supported are MP3, WAV, AAC, FLAC, AMR, 3GP, M4A, WMA, OGG and APE. Audio bitrate: 128–256 Kbps. Audio size: The file must be below 600 MB. The maximum duration is 3 hours. Audio file content languages: Mandarin, English and Cantonese are supported. When a video file is input, it supports separating the audio track from the video file and independent moderation of the audio content. |
Text | Encoding in Base64 should be performed first. The length of the original text must not exceed 10,000 UTF-8 encoded characters. HTML and TXT formats are supported. Encoding in UTF8 and GBK is supported. The size of files can not exceed 1 MB. Chinese, English, and Arabic numerals are supported. |
Live stream | Live stream duration: less than 5 hours. Live stream protocols: RTMP, HLS, HTTP, HTTPS and other major protocols. Live stream resolution: up to 1920x1080 (1080p). Default concurrency moderation route limit: 10 routes. If the number of concurrency moderation routes are exceeded, the interface will return an error. |
Document | Documents are converted to images for moderation through the document processing service. Currently, dozens of document formats such as PDF, PPT, and Excel are supported. For more information, see Document Preview Overview. |
Webpage | The system can automatically detect webpage files and recognize non-compliant content in OCR, object detection (such as object, advertising logo, and QR code), and image recognition dimensions based on the deep learning technology. |
Region | Code | |
Chinese mainland | Beijing | ap-beijing |
| Nanjing | ap-nanjing |
| Shanghai | ap-shanghai |
| Guangzhou | ap-guangzhou |
| Chengdu | ap-chengdu |
| Chongqing | ap-chongqing |
Hong Kong (China) and overseas | Hong Kong (China) | ap-hongkong |
| Singapore | ap-singapore |
| Frankfurt | eu-frankfurt |
Was this page helpful?