Duplicate from sy000/ChatGarmentDataset

Co-authored-by: Siyuan Bian <sy000@users.noreply.huggingface.co>

.gitattributes

@@ -0,0 +1,83 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.lz4 filter=lfs diff=lfs merge=lfs -text
+*.mds filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text
+# Audio files - uncompressed
+*.pcm filter=lfs diff=lfs merge=lfs -text
+*.sam filter=lfs diff=lfs merge=lfs -text
+*.raw filter=lfs diff=lfs merge=lfs -text
+# Audio files - compressed
+*.aac filter=lfs diff=lfs merge=lfs -text
+*.flac filter=lfs diff=lfs merge=lfs -text
+*.mp3 filter=lfs diff=lfs merge=lfs -text
+*.ogg filter=lfs diff=lfs merge=lfs -text
+*.wav filter=lfs diff=lfs merge=lfs -text
+# Image files - uncompressed
+*.bmp filter=lfs diff=lfs merge=lfs -text
+*.gif filter=lfs diff=lfs merge=lfs -text
+*.png filter=lfs diff=lfs merge=lfs -text
+*.tiff filter=lfs diff=lfs merge=lfs -text
+# Image files - compressed
+*.jpg filter=lfs diff=lfs merge=lfs -text
+*.jpeg filter=lfs diff=lfs merge=lfs -text
+*.webp filter=lfs diff=lfs merge=lfs -text
+# Video files - compressed
+*.mp4 filter=lfs diff=lfs merge=lfs -text
+*.webm filter=lfs diff=lfs merge=lfs -text
+garment_restpose_data_v1.json filter=lfs diff=lfs merge=lfs -text
+editing/random_newgarments_textsewing.json filter=lfs diff=lfs merge=lfs -text
+generation/data_detailtext_singlegarment_v2.json filter=lfs diff=lfs merge=lfs -text
+generation/data_detailtext_v2.json filter=lfs diff=lfs merge=lfs -text
+generation/data_detailtextimg_singlegarment_v2.json filter=lfs diff=lfs merge=lfs -text
+generation/data_detailtextimg_v2.json filter=lfs diff=lfs merge=lfs -text
+generation/data_detailtextimg_v3.json filter=lfs diff=lfs merge=lfs -text
+generation/data_detailtextimg_v4.json filter=lfs diff=lfs merge=lfs -text
+generation/data_img_v2.json filter=lfs diff=lfs merge=lfs -text
+generation/data_img_v3.json filter=lfs diff=lfs merge=lfs -text
+generation/data_restpose_img_v1.json filter=lfs diff=lfs merge=lfs -text
+shhq/sshq_llavaformat_list_detaildescp.json filter=lfs diff=lfs merge=lfs -text
+training/editing/random_newgarments_textsewing.json filter=lfs diff=lfs merge=lfs -text
+training/shhq/sshq_llavaformat_list_detaildescp.json filter=lfs diff=lfs merge=lfs -text
+training/synthetic/data_detailtext_singlegarment_v2.json filter=lfs diff=lfs merge=lfs -text
+training/synthetic/data_detailtext_v2.json filter=lfs diff=lfs merge=lfs -text
+training/synthetic/data_detailtextimg_singlegarment_v2.json filter=lfs diff=lfs merge=lfs -text
+training/synthetic/data_detailtextimg_v2.json filter=lfs diff=lfs merge=lfs -text
+training/synthetic/data_detailtextimg_v3.json filter=lfs diff=lfs merge=lfs -text
+training/synthetic/data_detailtextimg_v4.json filter=lfs diff=lfs merge=lfs -text
+training/synthetic/data_img_v2.json filter=lfs diff=lfs merge=lfs -text
+training/synthetic/data_img_v3.json filter=lfs diff=lfs merge=lfs -text
+training/synthetic/data_img_v4.json filter=lfs diff=lfs merge=lfs -text
+training/synthetic/data_restpose_img_v1.json filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,88 @@
+# ChatGarment Dataset
+
+## Introduction
+This repository provides the dataset for training and evaluating [ChatGarment](https://chatgarment.github.io/).
+
+## 1. Formatted Data
+These data are intended for general use.
+
+#### Garment Image Archives
+All garment images are stored in ``.zip`` files: ``garments_imgs_v1_*.zip``, ``garments_imgs_v2_*.zip``, ``garments_imgs_v3.zip``, ``garments_imgs_v4.zip``, etc. ``garments_imgs_v1_*.zip``, ``garments_imgs_v2_*.zip`` are split using the ``zipsplit`` command for ease of distribution.
+
+#### v1 & v2 Annotations
+We utilize the [GarmentCodeData v1](https://www.research-collection.ethz.ch/handle/20.500.11850/673889) dataset as the base garment dataset for garment sampling. The annotations are given in ``garment_restpose_data_v1.json`` and ``garment_posed_data_v2.json``.
+
+Here is the structure of the ``.json`` files:
+```json
+[{
+    'upper_garment': ... // Directory of upper garments in GarmentCodeData v1 dataset
+    'lower_garment': ... // Directory of lower garments in GarmentCodeData v1 dataset
+    'whole_garment': ... // Directory of wholebody garments in GarmentCodeData v1 dataset
+    'text_label': ... // Text descriptions for the outfit
+    'img_name': ... // Directory of outfit images in garments_imgs_v1 or garments_imgs_v2
+    'img_restpose_name': // Directory of A-pose outfit images in garments_imgs_v1 or garments_imgs_v2
+},
+...
+]
+```
+
+* **Dataset v1** contains only A-pose garments, and the text labels are extracted directly from ``dataset_properties_default_body.yaml`` in GarmentCodeData v1 dataset. 
+* **Dataset v2** contains garments in different poses. The text labels are generated by GPT-4o.
+
+
+#### v3 & v4 Annotations
+These versions are generated through customized sampling strategies. The annotations are provided in ``garment_posed_data_v3.json`` and ``garment_posed_data_v4.json``.
+
+Here is the structure of the ``.json`` files:
+```json
+[{
+    'upper_garment': ... // Directory of upper garments in garments_imgs_v4
+    'lower_garment': ... // Directory of lower garments in garments_imgs_v4
+    'whole_garment': ... // Directory of wholebody garments in garments_imgs_v4
+    'text_label_list': ... // A list of slightly different text descriptions for the outfit
+    'img_name': ... // Directory of outfit images in garments_imgs_v4
+},
+...
+]
+```
+**Dataset v3** and **Dataset v4** contain garments in different poses. The text labels are generated by GPT-4o.
+
+
+
+## 2. Training Data for ChatGarment
+We provide LLaVA-format JSON files used to train ChatGarment in the ``training`` folder.
+
+#### a. Synthetic Image and Text Dataset
+* Garments rendered in Blender.
+* Descriptions generated with GPT-4o.
+* Available at: training/synthetic/
+
+#### b. Synthetic Garment Editing Dataset
+* Available at: ``training/editing``.
+
+#### c. Text Labels for SHHQ Dataset
+* Available at: ``training/shhq``.
+
+
+## 3. Evaluation Data for ChatGarment
+* CloSE Evaluation: ``evaluations/close_eva_imgs.zip``
+* Dress4D Evaluation: ``evaluations/dress4d_eva_imgs.zip``
+* Garment Editing Evaluation: ``evaluations/garment_edit_eva.json``
+
+
+## 4. Extra Data
+The ``new_garments/`` folder contains:
+
+* NEW Rest-pose garments used in ChatGarment (excluding v1 and v2 garments from GarmentCodeData)
+* Corresponding ``.obj`` mesh files
+
+
+## Citation
+```bibtex
+@article{bian2024chatgarment,
+  title={ChatGarment: Garment Estimation, Generation and Editing via Large Language Models},
+  author={Bian, Siyuan and Xu, Chenghao and Xiu, Yuliang and Grigorev, Artur and Liu, Zhen and Lu, Cewu and Black, Michael J and Feng, Yao},
+  journal={arXiv preprint arXiv:2412.17811},
+  year={2024}
+} 
+```

evaluations/close_eva_imgs.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0c211d24e2e284d1f4ec87f2a0bfeecc4fd36a8593c8dfec51dfa3fa6d8cc886
+size 26671673

evaluations/dress4d_eva_imgs.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:af6b6471911000d36464ad9239066ed7ce2c3b8afe9a497693b868b1993c99b6
+size 75647310

garment_restpose_data_v1.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f689a4e4b8b23322350a6ffd1f414702c75dc9b104941bc33b5e591295697e32
+size 32613000

garments_imgs_v1_1.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a1d8e3166ce970fe7378fe7a4a36051a9d7dab1ccac0dff75014cb5cc21327d4
+size 32210405861

garments_imgs_v1_2.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eb61652ce31a0fa31982345cc466c3421cf14d809d75f0850aa9d3c618867648
+size 32209904921

garments_imgs_v1_3.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:90249d69b98f40502e8a9c6ed580037953332ae88680145eda78bf8b5f67e79f
+size 32209862431

garments_imgs_v1_4.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d81ae1f98fbca037e9087241cfef60c052db18a848f20c2b3a2e14e0880535d7
+size 32209902793

garments_imgs_v1_5.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b34f76bbdd35b0f4505125ea2024bcd98555b085d57ae7fa521412a17a9d1501
+size 28076526234

garments_imgs_v2_1.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:42c17115a271ea6b594906878c153167164e755597010e3c606bbe631bca80ec
+size 32209650967

garments_imgs_v2_2.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4c9584bf694a9d551a27a82699e7748fb6db24af071bd534cd175265ad3cedfb
+size 32209083485

garments_imgs_v2_3.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b072c70cf4048a99e47b0c3073c3518ba481ad91e0a009b238c93f4ee414c331
+size 8118467898

garments_imgs_v3.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:253d23251bd888b0a0b14e69528639e506293d9d497478009ae5993b8248b244
+size 26768738535

garments_imgs_v4.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dae196c96da8d9b3f81ce50c401ae4eb5ccdf1cef0d89a24ca001331179a1cb6
+size 19493405541

gpt4o_prompts.txt

@@ -0,0 +1,27 @@
+I will provide an image of a human model wearing several garments. Describe the outer layer garments the model is wearing. In the image, the model may wear one upper garment and one lower garment, or the model may wear a single wholebody garment. Avoid describing extra accessories such as the scarves, socks, watch, badges, and etc.
+
+For each garment, you should generate TWO strings.
+
+In the first string, describe the garment type (If THE SUBJECT HAS A NAME, INCLUDE ITS NAME FIRST!);
+
+  Example phrases for the first string: "hood", "T-shirt", "jacket", "tuxedo", etc.
+
+
+In the second string, describe the overall global geometric features of the garment (DO NOT INCLUDE ANY INFO ABOUT THE HUMAN MODEL AND THE COLOR INFO OF THE GARMENT) using several different short phrases split by ',' with the following tips: 
+
+  Example rules:
+  	Describe the length of the sleeves: long, normal, short, sleeveless, etc.
+  	Describe if it has a hood: with a hood, etc.
+  	Describe the length of the dress: long, normal, short, etc.
+  	Describe the width of the garment: wide, normal, narrow, etc.
+    Describe the length of the garment: long, normal, short, etc.
+  	Describe the length of the legs of trousers: long, normal, short, etc.
+
+  Please follow the example rules above (not limited to these examples) to describe the geometric features of the garment.
+
+  Example phrases for the second string: "long sleeves", "wide garment", "with a hood", "deep collar", "sleeveless"...
+
+
+Please strictly avoid mentioning color, texture, and material.
+
+In the image, if the model is wearing one upper garment and one lower garment, return the results in the following format: {"upper garment": [upper garment type, upper garment geometric features], "lower garment": [lower garment type, lower garment geometric features]}. Otherwise, the model is wearing a single wholebody garment , return the results in the following format: {"wholebody garment": [wholebody garment type, wholebody garment geometric features]}. Return only the JSON dictionary in the above format with a length of 1 or 2.

new_garments/garment_v3_1.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:09c61cc3fd050a94ba7da82d60a2771e78009e04a2307e71b6b84cb412cd0f59
+size 42946956653

new_garments/garment_v3_2.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8da5c255b996dba7413ae68b05e8b22cf9c04d179a5565c806302669bcba2bb3
+size 27045293179

new_garments/garment_v4.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0700ded8761825674877c6a0da474d2b602dbd8a0329fda2c7ce8ff9fa7c94f8
+size 38235534425

training/editing/random_newgarments_textsewing.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:069ecd4bd535d4aa468f7d625d41788a07f7e201e1e1ae40ada2ded5dde93391
+size 20461564

training/shhq/sshq_llavaformat_list_detaildescp.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b967a9a49c545d0562ef3e2f33e556ec8049b7bcf13883fc907c1521305a424d
+size 25326654

training/synthetic/data_detailtext_singlegarment_v2.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:96252333e8af1dd0ec9976c4a663240920bf978fe7e1b577304968b4c69a5b79
+size 22170245

training/synthetic/data_detailtext_v2.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:221bab31b14282303376b3ff3fe669290334dea368af5232619aa40535ef6f24
+size 21569219

training/synthetic/data_detailtextimg_singlegarment_v2.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:086a0a53eeb3e44ef5267fc7cf015f5e2be340f7ea546ae256eee2c3ad1ab789
+size 865401551

training/synthetic/data_detailtextimg_v2.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4566604249f030930d2ba38777068ba94fa5b340595318a84fbcf8d2db40a3fe
+size 813358152

training/synthetic/data_detailtextimg_v3.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a2b8e6d4865ffa3d57c4f507b91920bcdb0c727bfc08b3a9e7c7090530bf3c6b
+size 234929756

training/synthetic/data_detailtextimg_v4.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:176eacf7ded179367dbb4f98b609547420f7bae008ce897756f998526cf39e0a
+size 166010466

training/synthetic/data_img_v2.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:317bb871d39092409fc488e39f245314fb63a89867f8050e5f65799d551f2333
+size 734434620

training/synthetic/data_img_v3.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7b1e4cfd0c84e1f441afcacada3eb85a8aab89b77f0d2a208064298f97637f2e
+size 759062756

training/synthetic/data_img_v4.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f7b457235d9c16cb48db83aea537ec0ff7a03da61487e4ea3259699de33a085e
+size 128112640

training/synthetic/data_restpose_img_v1.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:37e1b9e09ffec28eb09bf235db8d7bda7b0c187b2646f5d0c362744e0bb75e51
+size 1408155834