Document upstream media hydration workflow

README.md

@@ -288,20 +288,30 @@ WHERE v.row_id = 100;
 ## Media downloaded from upstream sources
 
 Media from the following datasets must be obtained from the original source.
-Their annotations and source-image mappings are included here.
+This repository includes their annotations, JSONL indexes, source-image
+mappings, and metadataset configuration, but not their image files.
 
-| Dataset folder | Media source |
-|---|---|
-| `crowdhuman` | [CrowdHuman](https://www.crowdhuman.org/download.html) |
-| `deepfashion2` | [DeepFashion2](https://github.com/switchablenorms/DeepFashion2) |
-| `flickr30k` | [Flickr30K](https://shannon.cs.illinois.edu/DenotationGraph/data/index.html) |
-| `partimagenet` | [ImageNet](https://www.image-net.org/download.php) |
-| `object365` | [Objects365](https://www.objects365.org/download.html) |
-| `sku110k` | [SKU-110K](https://github.com/eg4000/SKU110K_CVPR19) |
-| `unsplash` | [Unsplash Dataset](https://unsplash.com/data) |
-
-After downloading a source dataset with its original directory and filenames,
-build the local media shards from the public mapping:
+| Dataset folder | Download images from | Expected path under `--source-root` |
+|---|---|---|
+| `crowdhuman` | [CrowdHuman](https://www.crowdhuman.org/download.html) | `train/Images/<image>.jpg` |
+| `deepfashion2` | [DeepFashion2](https://github.com/switchablenorms/DeepFashion2) | `train/image/<six-digit-id>.jpg` |
+| `flickr30k` | [Flickr30K](https://shannon.cs.illinois.edu/DenotationGraph/data/index.html) | `flickr30k-images/<flickr-id>.jpg` |
+| `partimagenet` | [ImageNet](https://www.image-net.org/download.php) | `train/<wnid>/<wnid>_<id>.JPEG` |
+| `object365` | [Objects365](https://www.objects365.org/download.html) | `images/train/patch*/objects365_v1_<id>.jpg` |
+| `sku110k` | [SKU-110K](https://github.com/eg4000/SKU110K_CVPR19) | `images/train_<id>.jpg` |
+| `unsplash` | [Unsplash Dataset](https://unsplash.com/data) | `raw_unsplash_images/<photo-id>.jpg` |
+
+`metadataset.yaml` reads the datasets whose images are already hosted in this
+repository. Use `metadataset-full.yaml` only after all media listed above has
+been prepared locally.
+
+### Prepare an upstream media dataset
+
+1. Download the images from the official source and accept its terms.
+2. Keep the upstream filenames and arrange the files under a source root using
+   the layout shown above.
+3. Install `megatron-energon==7.4.0`.
+4. Run the hydration tool with the matching dataset ID and source root.
 
 ```bash
 python tools/hydrate_restricted_media.py \
@@ -310,8 +320,24 @@ python tools/hydrate_restricted_media.py \
   --source-root /data/upstream/CrowdHuman
 ```
 
-The tool verifies mapped files and creates the TAR indexes and Energon metadata
-expected by `metadataset-full.yaml`.
+Run this once for each of the seven dataset folders, changing `--dataset` and
+`--source-root` each time. The accepted dataset IDs are:
+
+```text
+crowdhuman
+deepfashion2
+flickr30k
+partimagenet
+object365
+sku110k
+unsplash
+```
+
+The tool uses `mappings/media/<pool_id>/part-*.parquet` as the file inventory.
+It resolves every `source_relative_path` under `--source-root`, verifies the
+file, and builds the local TAR indexes and Energon metadata required by
+`metadataset-full.yaml`. The tool prepares local media only; upstream downloads
+must be completed separately.
 
 ## License Information
 

README_CN.md

@@ -276,21 +276,28 @@ WHERE v.row_id = 100;
 
 ## 需要从上游下载的媒体
 
-以下数据集的媒体需要从原始来源获取；对应的标注和 source-image mapping 已经
-包含在本仓库中。
+以下数据集的图片需要从原始来源获取。本仓库已经包含它们的标注、JSONL 索引、
+source-image mapping 和 metadataset 配置，但不包含图片文件。
 
-| 数据集目录 | 媒体来源 |
-|---|---|
-| `crowdhuman` | [CrowdHuman](https://www.crowdhuman.org/download.html) |
-| `deepfashion2` | [DeepFashion2](https://github.com/switchablenorms/DeepFashion2) |
-| `flickr30k` | [Flickr30K](https://shannon.cs.illinois.edu/DenotationGraph/data/index.html) |
-| `partimagenet` | [ImageNet](https://www.image-net.org/download.php) |
-| `object365` | [Objects365](https://www.objects365.org/download.html) |
-| `sku110k` | [SKU-110K](https://github.com/eg4000/SKU110K_CVPR19) |
-| `unsplash` | [Unsplash Dataset](https://unsplash.com/data) |
+| 数据集目录 | 图片下载入口 | `--source-root` 下的预期路径 |
+|---|---|---|
+| `crowdhuman` | [CrowdHuman](https://www.crowdhuman.org/download.html) | `train/Images/<image>.jpg` |
+| `deepfashion2` | [DeepFashion2](https://github.com/switchablenorms/DeepFashion2) | `train/image/<six-digit-id>.jpg` |
+| `flickr30k` | [Flickr30K](https://shannon.cs.illinois.edu/DenotationGraph/data/index.html) | `flickr30k-images/<flickr-id>.jpg` |
+| `partimagenet` | [ImageNet](https://www.image-net.org/download.php) | `train/<wnid>/<wnid>_<id>.JPEG` |
+| `object365` | [Objects365](https://www.objects365.org/download.html) | `images/train/patch*/objects365_v1_<id>.jpg` |
+| `sku110k` | [SKU-110K](https://github.com/eg4000/SKU110K_CVPR19) | `images/train_<id>.jpg` |
+| `unsplash` | [Unsplash Dataset](https://unsplash.com/data) | `raw_unsplash_images/<photo-id>.jpg` |
+
+`metadataset.yaml` 读取图片已经包含在本仓库中的数据集。只有在本地补齐上述
+全部媒体后，才使用 `metadataset-full.yaml`。
+
+### 准备一个上游媒体数据集
 
-下载上游数据并保留原始目录和文件名后，可以根据公开 mapping 构建本地媒体
-shard：
+1. 从官方入口下载图片并接受相应条款。
+2. 保留上游文件名，并按照上表结构放到一个 source root 下。
+3. 安装 `megatron-energon==7.4.0`。
+4. 使用对应的数据集 ID 和 source root 运行 hydration 工具。
 
 ```bash
 python tools/hydrate_restricted_media.py \
@@ -299,8 +306,23 @@ python tools/hydrate_restricted_media.py \
   --source-root /data/upstream/CrowdHuman
 ```
 
-工具会校验 mapping 引用的文件，并生成 `metadataset-full.yaml` 所需的 TAR
-索引和 Energon 元数据。
+对七个数据集目录分别运行一次，并相应修改 `--dataset` 和 `--source-root`。
+可用的数据集 ID 为：
+
+```text
+crowdhuman
+deepfashion2
+flickr30k
+partimagenet
+object365
+sku110k
+unsplash
+```
+
+工具使用 `mappings/media/<pool_id>/part-*.parquet` 作为文件清单，在
+`--source-root` 下解析每个 `source_relative_path`，校验文件，并生成
+`metadataset-full.yaml` 所需的本地 TAR 索引和 Energon 元数据。该工具只负责
+准备本地媒体，上游图片需要预先自行下载。
 
 ## License Information
 

manifests/files.jsonl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2a3b2b07a2d6574a9500be5b9ecfded0e512342a14670dc6bc7488aa3c24dd0e
+oid sha256:bd852c49255aae32ef41bb5152bddc6f29fd0b5ecb5e14389e2f3a130efd9f7a
 size 295285