mirror of
https://github.com/Tavish9/any4lerobot.git
synced 2026-05-20 08:29:40 +00:00
Jibby Nguyen
ef184e44be
* Support robocasa2lerobot * Support robocasa2lerobot * NIT: formatting * update to latest lerobot * update readme * Apply suggestion from @gemini-code-assist[bot] Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com> * fix h5py open --------- Co-authored-by: Tavish <tavish9.chen@gmail.com> Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
71 lines
2.2 KiB
Plaintext
71 lines
2.2 KiB
Plaintext
{
|
|
"cells": [
|
|
{
|
|
"cell_type": "markdown",
|
|
"id": "44b6da09",
|
|
"metadata": {},
|
|
"source": [
|
|
"# Extract subset data \n",
|
|
"\n",
|
|
"Original hdf5 file contains about 3000 episodes. However, it contains a key \"masks\", which contain list of subset demo_ids. For example: 30_demos : [demo123, demo234, demo 345, etc.]\n",
|
|
"\n",
|
|
"Run the code bellow to extract only chosen subset demos, which is much smaller and easier for later process."
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"id": "6ac64550",
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"import h5py\n",
|
|
"\n",
|
|
"DATA_DIR=\"direction/to/your/hdf5/files/\"\n",
|
|
"# E.x: DATA_DIR=\"/projects/extern/kisski/kisski-spath/dir.project/VLA_3D/binh/robocasa/test\"\n",
|
|
"\n",
|
|
"# file_name = \"PnPCabToCounter.hdf5\"\n",
|
|
"# file_name = \"PnPCounterToCab.hdf5\"\n",
|
|
"# file_name = \"CoffeeSetupMug.hdf5\"\n",
|
|
"# file_name = \"TurnOnMicrowave.hdf5\"\n",
|
|
"file_name = \"TurnOffStove.hdf5\"\n",
|
|
"\n",
|
|
"file_path = DATA_DIR + \"/\" + file_name\n",
|
|
"\n",
|
|
"f = h5py.File(file_path, 'r')\n",
|
|
"chosen_demo_list = []\n",
|
|
"for i in f['mask']['100_demos'][:]: # or \"30_demos\"\n",
|
|
" chosen_demo_list.append(i.decode('utf-8'))\n",
|
|
" \n",
|
|
"chosen_data = []\n",
|
|
"for k in f['data'].keys():\n",
|
|
" if k in chosen_demo_list:\n",
|
|
" chosen_data.append(f['data'][k])\n",
|
|
" \n",
|
|
"with h5py.File(f\"direction_to_your_new_extracted_subset/{file_name}\", \"w\") as out:\n",
|
|
" out_data = out.create_group(\"data\")\n",
|
|
" \n",
|
|
" for key, val in f['data'].attrs.items():\n",
|
|
" out_data.attrs[key] = val # IMPORTANT: set attributes for new hdf5 files (need for reset env and later re-render)\n",
|
|
"\n",
|
|
" for grp in chosen_data:\n",
|
|
" name = grp.name.split(\"/\")[-1] # demo_xxx\n",
|
|
" grp.file.copy(grp, out_data, name=name)"
|
|
]
|
|
}
|
|
],
|
|
"metadata": {
|
|
"kernelspec": {
|
|
"display_name": "robocasa",
|
|
"language": "python",
|
|
"name": "python3"
|
|
},
|
|
"language_info": {
|
|
"name": "python",
|
|
"version": "3.10.19"
|
|
}
|
|
},
|
|
"nbformat": 4,
|
|
"nbformat_minor": 5
|
|
}
|