人體解析（Human Parse）開源資料集整理

文章目錄

• 背景
• PASCAL-Person-Part
• ATR
• LIP
• MHP
• CIHP
• 總結

背景

Human Parse指将在圖像中的人分割為多個語義上一緻的區域。

目前該領域主要有以下開源資料集：PASCAL-Person-Part、ATR、LIP、MHP、CIHP。

PASCAL-Person-Part

PASCAL-Person-Part在PSACAL VOC2010資料集基礎上增加分割标注，該資料集存在多人場景，其中訓練集1716張，測試集1817張，共标注7個類别：‘Background’, ‘Head’, ‘Torso’, ‘Upper Arms’, ‘Lower Arms’, ‘Upper Legs’, ‘Lower Legs’。如下圖所示。

下載下傳連結

論文：《Detect What You Can: Detecting and Representing Objects using Holistic Models and Body Parts》

ATR

ATR(Active Template Regression)資料集為單人場景人體解析資料集，該資料集共17700張，其中訓練集16000張，驗證集700張，測試集1000張。共标注18個類别：‘Background’, ‘Hat’, ‘Hair’, ‘Sunglasses’, ‘Upper-clothes’, ‘Skirt’, ‘Pants’, ‘Dress’, ‘Belt’, ‘Left-shoe’, ‘Right-shoe’, ‘Face’, ‘Left-leg’, ‘Right-leg’, ‘Left-arm’, ‘Right-arm’, ‘Bag’, ‘Scarf’.如下圖所示。

下載下傳連結

論文：《Deep Human Parsing with Active Template Regression》

LIP

LIP(Look into Person)資料集為單人場景人體解析資料集，共50462張，其中訓練集30462張，驗證集10000張，測試集10000張，共标注20個類别：‘Background’, ‘Hat’, ‘Hair’, ‘Glove’, ‘Sunglasses’, ‘Upper-clothes’, ‘Dress’, ‘Coat’, ‘Socks’, ‘Pants’, ‘Jumpsuits’, ‘Scarf’, ‘Skirt’, ‘Face’, ‘Left-arm’, ‘Right-arm’, ‘Left-leg’, ‘Right-leg’, ‘Left-shoe’, ‘Right-shoe’.如下圖所示。

下載下傳連結

論文： 《Look into Person: Self-supervised Structure-sensitive Learning and A New Benchmark for Human Parsing》

MHP

MHP（Multi-Human-Parsing）資料集為多人場景人體解析資料集。MHP v1.0包括4980張圖檔，平均每張圖包括3人，訓練集3000張，驗證集1000張，測試集980張，共标注19個類别：“background”, “hat”, “hair”, “sunglasses”, “upper clothes”, “skirt”, “pants”, “dress”, “belt”, “left shoe”, “right shoe”, “face”, “left leg”, “right leg”, “left arm”, “right arm”, “bag”, “scarf” and “torso skin”，如下圖所示：

下載下傳位址：baidu drive (password: cmtp).

MHP v2.0包括25403圖檔，訓練集15403張，驗證集5000張，測試集5000張，共标注59個類别： “background”, “cap/hat”, “helmet”, “face”, “hair”, “left-arm”, “right-arm”, “left-hand”, “right-hand”, “protector”, “bikini/bra”, “jacket/windbreaker/hoodie”, “t-shirt”, “polo-shirt”, “sweater”, “singlet”, “torso-skin”, “pants”, “shorts/swim-shorts”, “skirt”, “stockings”, “socks”, “left-boot”, “right-boot”, “left-shoe”, “right-shoe”, “left-highheel”, “right-highheel”, “left-sandal”, “right-sandal”, “left-leg”, “right-leg”, “left-foot”, “right-foot”, “coat”, “dress”, “robe”, “jumpsuit”, “other-full-body-clothes”, “headwear”, “backpack”, “ball”, “bats”, “belt”, “bottle”, “carrybag”, “cases”, “sunglasses”, “eyewear”, “glove”, “scarf”, “umbrella”, “wallet/purse”, “watch”, “wristband”, “tie”, “other-accessary”, “other-upper-body-clothes” and “other-lower-body-clothes”.如下圖所示，

下載下傳位址：google drive

github：https://github.com/ZhaoJ9014/Multi-Human-Parsing

CIHP

CIHP（Crowd Instance-level Human Parsing）資料集為多人場景人體解析資料集，共38280張，其中訓練集28280張，驗證集5000張，測試集5000張，共标注20個類别：Background, Hat, Hair, Sunglasses, Upper-clothes, Dress, Coat, Socks, Pants, Gloves, Scarf, Skirt, Torsoskin, Face, Right/Left arm, Right/Left leg, and Right/Left shoe.如下圖所示，

下載下傳位址

論文：《Instance-level Human Parsing via Part Grouping Network》

總結

資料集	場景	訓練集	驗證集	測試集	類别	下載下傳連結
PASCAL-Person-Part	多人	1716	-	1817	7	下載下傳連結
ATR	單人	16000	700	1000	18	-
LIP	單人	30462	10000	10000	20	下載下傳連結
MHP v1.0	多人	3000	1000	980	19	baidu drive (password: cmtp)
MHP v2.0	多人	15403	5000	5000	59	google drive
CIHP	多人	28280	5000	5000	20	下載下傳位址