Structures for BOORU CHARS 2024 (tab separated) supplementary metadata : BC_2024.tsv - file / image info 1.260.629 rows * general BOORU - imageboard FID - post id SPATH - zip-archive locator RNK - quality rank inside AR+headcount volume FNAME - file name "%website% - %id% - %up_to_3_copyrights% ~ %up_to_5_characters% (%up_to_2_artists%).jpg" * sample inside this torrent S_FSIZE - file size in bytes S_ISIZE - image size WIDTHxHEIGHT S_JQ - JPEG quality estimation (NOTE value 81 is equivalent cjpegli -q 96) * grabbed image G_DATE - posting date G_RATN - rating G_SCORE - popularity @ booru G_ISIZE - WIDTHxHEIGHT @ booru G_EXT - original extension G_FSIZE - bytes @ booru G_JQ - JPEG quality estimation just grabbed JPEGs G_FMD5 - MD5 @ booru or just grabbed * calculated by Image Magick TENTR - enthropy ENT_R - rank by enthropy TSKEW - skewness SKEW_R - rank by skewness TSTDDEV - standard deviation SDEV_R - rank by standard deviation TCOLORS - colors count COLR_R - rank by colors count MEANG - mean grey channel GREY_R - rank by mean grey MAXIMAG - maximum in grey channel BOUNDBOX - frame by ImageMagick BB_R - rank by frame-to-image size * advanced detection results TXSIZE - relative text-to-image size in percents TXS_R - rank by text/image size TXCNT - count of text pieces TXN_R - rank by count of text pieces YHEAD - heads count YHMAX - max head/image size YHAVG - average head/image size YH_R - rank by crowdeness * tags count by category TAGS_C - (copyright) TAGS_P - (character) TAGS_A - (artist) TAGS_G - (general) TAGS_X - (technical, others) BC_2024_tags.tsv - tags list enriched 49.041.220 rows booru + fid - imageboard name and ID, refer to BC_2024 TAG - text tag TAG_ID - numeric tag as of DANBOORU TAG_CAT - tag category as of DANBOORU (3=copyright, 4=character, 1=artist, 0=general, 5=others) DANB_TAG - DANBOORU text tag * for CAT in (0,1,5) DANBOORU-synonym for TAG if != TAG * for CAT in (3,4) DANBOORU-franchise (generalised copyright series) for TAG if != TAG BC_2024_yolo.tsv - detailed data about torso components detections 4.431.887 rows booru + fid - imageboard name and ID, refer to BC_2024 obj - object short name [for custom model](https://github.com/aperveyev/booru_yolo/tree/main/models) prob - detection probability x, y, w, h - bounbox in YOLO format (relative, centered) oid - object ID (some hash) unique within image oup - "upper" oid (head for bust and wing, bust for belly) odn - "down" oid (bust and wing for head, belly for bust) BC_2024_yolov11m_aa22.pt [Ultralitics YOLOv11](https://github.com/ultralytics/ultralytics) object detection model The whole evolved BOORU CHARS project consists of (in release order): - [BOORU_CHARS_2021](https://nyaa.si/view/1384820) 1.593.429 images 472 GB topic starter - [BOORU_CHARS_2015](https://nyaa.si/view/1468367) 463.873 images 148 GB old stuff - [BOORU_CHARS_2022](https://nyaa.si/view/1547662) 705.467 images 191 GB newcomers - [BOORU_CHARS_2023](https://nyaa.si/view/1740396) 1.153.513 images 302 GB FINAL for 1280/1024px