зеркало из https://github.com/microsoft/torchgeo.git
Cleaner benchmark header levels
This commit is contained in:
Adam J. Stewart
Родитель
d7d901c866
Коммит
b60948e598
|
|
@ -1,243 +1,6 @@
|
|||
{
|
||||
"cells": [
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"source": [
|
||||
"### Imports"
|
||||
],
|
||||
"metadata": {
|
||||
"nteract": {
|
||||
"transient": {
|
||||
"deleting": false
|
||||
}
|
||||
}
|
||||
}
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"source": [
|
||||
"import os\r\n",
|
||||
"import sys\r\n",
|
||||
"import time\r\n",
|
||||
"from typing import Tuple\r\n",
|
||||
"\r\n",
|
||||
"sys.path.append(\"../..\")\r\n",
|
||||
"\r\n",
|
||||
"from torch.utils.data import DataLoader\r\n",
|
||||
"\r\n",
|
||||
"from torchgeo.datasets import ChesapeakeDE, NAIP\r\n",
|
||||
"from torchgeo.samplers import RandomGeoSampler, GridGeoSampler, RandomBatchGeoSampler\r\n",
|
||||
"\r\n",
|
||||
"\r\n",
|
||||
"ROOT = \"/mnt/blobfuse/adam-scratch\""
|
||||
],
|
||||
"outputs": [],
|
||||
"execution_count": 1,
|
||||
"metadata": {
|
||||
"gather": {
|
||||
"logged": 1629238744113
|
||||
}
|
||||
}
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"source": [
|
||||
"### Timing function"
|
||||
],
|
||||
"metadata": {
|
||||
"nteract": {
|
||||
"transient": {
|
||||
"deleting": false
|
||||
}
|
||||
}
|
||||
}
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"source": [
|
||||
"def time_epoch(dataloader: DataLoader) -> Tuple[float, int]:\r\n",
|
||||
" tic = time.time()\r\n",
|
||||
" i = 0\r\n",
|
||||
" for _ in dataloader:\r\n",
|
||||
" i += 1\r\n",
|
||||
" toc = time.time()\r\n",
|
||||
" return toc - tic, i"
|
||||
],
|
||||
"outputs": [],
|
||||
"execution_count": 2,
|
||||
"metadata": {
|
||||
"collapsed": true,
|
||||
"jupyter": {
|
||||
"source_hidden": false,
|
||||
"outputs_hidden": false
|
||||
},
|
||||
"nteract": {
|
||||
"transient": {
|
||||
"deleting": false
|
||||
}
|
||||
},
|
||||
"gather": {
|
||||
"logged": 1629238744228
|
||||
}
|
||||
}
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"source": [
|
||||
"### RandomGeoSampler"
|
||||
],
|
||||
"metadata": {
|
||||
"nteract": {
|
||||
"transient": {
|
||||
"deleting": false
|
||||
}
|
||||
}
|
||||
}
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"source": [
|
||||
"for cache in [False, True]:\r\n",
|
||||
" chesapeake = ChesapeakeDE(os.path.join(ROOT, \"chesapeake\", \"DE\"), cache=cache)\r\n",
|
||||
" naip = NAIP(os.path.join(ROOT, \"naip\"), crs=chesapeake.crs, res=chesapeake.res, cache=cache)\r\n",
|
||||
" dataset = chesapeake + naip\r\n",
|
||||
" sampler = RandomGeoSampler(naip.index, size=1000, length=888)\r\n",
|
||||
" dataloader = DataLoader(dataset, batch_size=12, sampler=sampler)\r\n",
|
||||
" duration, count = time_epoch(dataloader)\r\n",
|
||||
" print(duration, count)"
|
||||
],
|
||||
"outputs": [
|
||||
{
|
||||
"output_type": "stream",
|
||||
"name": "stdout",
|
||||
"text": [
|
||||
"296.582683801651 74\n",
|
||||
"54.20210099220276 74\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"execution_count": 8,
|
||||
"metadata": {
|
||||
"collapsed": true,
|
||||
"jupyter": {
|
||||
"source_hidden": false,
|
||||
"outputs_hidden": false
|
||||
},
|
||||
"nteract": {
|
||||
"transient": {
|
||||
"deleting": false
|
||||
}
|
||||
},
|
||||
"gather": {
|
||||
"logged": 1629248963725
|
||||
}
|
||||
}
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"source": [
|
||||
"### GridGeoSampler"
|
||||
],
|
||||
"metadata": {
|
||||
"nteract": {
|
||||
"transient": {
|
||||
"deleting": false
|
||||
}
|
||||
}
|
||||
}
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"source": [
|
||||
"for cache in [False, True]:\r\n",
|
||||
" chesapeake = ChesapeakeDE(os.path.join(ROOT, \"chesapeake\", \"DE\"), cache=cache)\r\n",
|
||||
" naip = NAIP(os.path.join(ROOT, \"naip\"), crs=chesapeake.crs, res=chesapeake.res, cache=cache)\r\n",
|
||||
" dataset = chesapeake + naip\r\n",
|
||||
" sampler = GridGeoSampler(naip.index, size=1000, stride=500)\r\n",
|
||||
" dataloader = DataLoader(dataset, batch_size=12, sampler=sampler)\r\n",
|
||||
" duration, count = time_epoch(dataloader)\r\n",
|
||||
" print(duration, count)"
|
||||
],
|
||||
"outputs": [
|
||||
{
|
||||
"output_type": "stream",
|
||||
"name": "stdout",
|
||||
"text": [
|
||||
"391.90197944641113 74\n",
|
||||
"118.0611424446106 74\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"execution_count": 4,
|
||||
"metadata": {
|
||||
"collapsed": true,
|
||||
"jupyter": {
|
||||
"source_hidden": false,
|
||||
"outputs_hidden": false
|
||||
},
|
||||
"nteract": {
|
||||
"transient": {
|
||||
"deleting": false
|
||||
}
|
||||
},
|
||||
"gather": {
|
||||
"logged": 1629239313388
|
||||
}
|
||||
}
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"source": [
|
||||
"### RandomBatchGeoSampler"
|
||||
],
|
||||
"metadata": {
|
||||
"nteract": {
|
||||
"transient": {
|
||||
"deleting": false
|
||||
}
|
||||
}
|
||||
}
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"source": [
|
||||
"for cache in [False, True]:\r\n",
|
||||
" chesapeake = ChesapeakeDE(os.path.join(ROOT, \"chesapeake\", \"DE\"), cache=cache)\r\n",
|
||||
" naip = NAIP(os.path.join(ROOT, \"naip\"), crs=chesapeake.crs, res=chesapeake.res, cache=cache)\r\n",
|
||||
" dataset = chesapeake + naip\r\n",
|
||||
" sampler = RandomBatchGeoSampler(naip.index, size=1000, batch_size=12, length=888)\r\n",
|
||||
" dataloader = DataLoader(dataset, batch_sampler=sampler)\r\n",
|
||||
" duration, count = time_epoch(dataloader)\r\n",
|
||||
" print(duration, count)"
|
||||
],
|
||||
"outputs": [
|
||||
{
|
||||
"output_type": "stream",
|
||||
"name": "stdout",
|
||||
"text": [
|
||||
"230.51380324363708 74\n",
|
||||
"53.99923872947693 74\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"execution_count": 10,
|
||||
"metadata": {
|
||||
"collapsed": true,
|
||||
"jupyter": {
|
||||
"source_hidden": false,
|
||||
"outputs_hidden": false
|
||||
},
|
||||
"nteract": {
|
||||
"transient": {
|
||||
"deleting": false
|
||||
}
|
||||
},
|
||||
"gather": {
|
||||
"logged": 1629249843438
|
||||
}
|
||||
}
|
||||
}
|
||||
],
|
||||
"nbformat": 4,
|
||||
"nbformat_minor": 0,
|
||||
"metadata": {
|
||||
"kernelspec": {
|
||||
"name": "python38-azureml",
|
||||
|
|
@ -261,8 +24,274 @@
|
|||
},
|
||||
"nteract": {
|
||||
"version": "nteract-front-end@1.0.0"
|
||||
},
|
||||
"colab": {
|
||||
"name": "benchmarking.ipynb",
|
||||
"provenance": [],
|
||||
"collapsed_sections": []
|
||||
}
|
||||
},
|
||||
"nbformat": 4,
|
||||
"nbformat_minor": 2
|
||||
"cells": [
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {
|
||||
"id": "OFXtoHmJClRf"
|
||||
},
|
||||
"source": [
|
||||
"# Benchmarking\n",
|
||||
"\n",
|
||||
"This tutorial benchmarks the performance of various sampling strategies, with and without caching."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {
|
||||
"nteract": {
|
||||
"transient": {
|
||||
"deleting": false
|
||||
}
|
||||
},
|
||||
"id": "hC3pauOLChi4"
|
||||
},
|
||||
"source": [
|
||||
"## Imports"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"metadata": {
|
||||
"gather": {
|
||||
"logged": 1629238744113
|
||||
},
|
||||
"id": "gjFiws-PChi8"
|
||||
},
|
||||
"source": [
|
||||
"import os\n",
|
||||
"import sys\n",
|
||||
"import time\n",
|
||||
"from typing import Tuple\n",
|
||||
"\n",
|
||||
"sys.path.append(\"../..\")\n",
|
||||
"\n",
|
||||
"from torch.utils.data import DataLoader\n",
|
||||
"\n",
|
||||
"from torchgeo.datasets import ChesapeakeDE, NAIP\n",
|
||||
"from torchgeo.samplers import RandomGeoSampler, GridGeoSampler, RandomBatchGeoSampler\n",
|
||||
"\n",
|
||||
"\n",
|
||||
"ROOT = \"/mnt/blobfuse/adam-scratch\""
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {
|
||||
"nteract": {
|
||||
"transient": {
|
||||
"deleting": false
|
||||
}
|
||||
},
|
||||
"id": "n6HwpMz7Chi-"
|
||||
},
|
||||
"source": [
|
||||
"## Timing function"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"metadata": {
|
||||
"collapsed": true,
|
||||
"jupyter": {
|
||||
"source_hidden": false,
|
||||
"outputs_hidden": false
|
||||
},
|
||||
"nteract": {
|
||||
"transient": {
|
||||
"deleting": false
|
||||
}
|
||||
},
|
||||
"gather": {
|
||||
"logged": 1629238744228
|
||||
},
|
||||
"id": "8-z6_y2xChi-"
|
||||
},
|
||||
"source": [
|
||||
"def time_epoch(dataloader: DataLoader) -> Tuple[float, int]:\n",
|
||||
" tic = time.time()\n",
|
||||
" i = 0\n",
|
||||
" for _ in dataloader:\n",
|
||||
" i += 1\n",
|
||||
" toc = time.time()\n",
|
||||
" return toc - tic, i"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": []
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {
|
||||
"nteract": {
|
||||
"transient": {
|
||||
"deleting": false
|
||||
}
|
||||
},
|
||||
"id": "I3pkKYoeChi_"
|
||||
},
|
||||
"source": [
|
||||
"## RandomGeoSampler"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"metadata": {
|
||||
"collapsed": true,
|
||||
"jupyter": {
|
||||
"source_hidden": false,
|
||||
"outputs_hidden": false
|
||||
},
|
||||
"nteract": {
|
||||
"transient": {
|
||||
"deleting": false
|
||||
}
|
||||
},
|
||||
"gather": {
|
||||
"logged": 1629248963725
|
||||
},
|
||||
"id": "jPjIZLF7Chi_",
|
||||
"outputId": "edcc8199-bd09-4832-e50c-7be8ac78995b"
|
||||
},
|
||||
"source": [
|
||||
"for cache in [False, True]:\n",
|
||||
" chesapeake = ChesapeakeDE(os.path.join(ROOT, \"chesapeake\", \"DE\"), cache=cache)\n",
|
||||
" naip = NAIP(os.path.join(ROOT, \"naip\"), crs=chesapeake.crs, res=chesapeake.res, cache=cache)\n",
|
||||
" dataset = chesapeake + naip\n",
|
||||
" sampler = RandomGeoSampler(naip.index, size=1000, length=888)\n",
|
||||
" dataloader = DataLoader(dataset, batch_size=12, sampler=sampler)\n",
|
||||
" duration, count = time_epoch(dataloader)\n",
|
||||
" print(duration, count)"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": [
|
||||
{
|
||||
"output_type": "stream",
|
||||
"name": "stdout",
|
||||
"text": [
|
||||
"296.582683801651 74\n",
|
||||
"54.20210099220276 74\n"
|
||||
]
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {
|
||||
"nteract": {
|
||||
"transient": {
|
||||
"deleting": false
|
||||
}
|
||||
},
|
||||
"id": "pHqLRDA_ChjB"
|
||||
},
|
||||
"source": [
|
||||
"## GridGeoSampler"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"metadata": {
|
||||
"collapsed": true,
|
||||
"jupyter": {
|
||||
"source_hidden": false,
|
||||
"outputs_hidden": false
|
||||
},
|
||||
"nteract": {
|
||||
"transient": {
|
||||
"deleting": false
|
||||
}
|
||||
},
|
||||
"gather": {
|
||||
"logged": 1629239313388
|
||||
},
|
||||
"id": "K67vnCK4ChjC",
|
||||
"outputId": "159ce99f-a438-4ecc-d218-9b9e28d02055"
|
||||
},
|
||||
"source": [
|
||||
"for cache in [False, True]:\n",
|
||||
" chesapeake = ChesapeakeDE(os.path.join(ROOT, \"chesapeake\", \"DE\"), cache=cache)\n",
|
||||
" naip = NAIP(os.path.join(ROOT, \"naip\"), crs=chesapeake.crs, res=chesapeake.res, cache=cache)\n",
|
||||
" dataset = chesapeake + naip\n",
|
||||
" sampler = GridGeoSampler(naip.index, size=1000, stride=500)\n",
|
||||
" dataloader = DataLoader(dataset, batch_size=12, sampler=sampler)\n",
|
||||
" duration, count = time_epoch(dataloader)\n",
|
||||
" print(duration, count)"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": [
|
||||
{
|
||||
"output_type": "stream",
|
||||
"name": "stdout",
|
||||
"text": [
|
||||
"391.90197944641113 74\n",
|
||||
"118.0611424446106 74\n"
|
||||
]
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {
|
||||
"nteract": {
|
||||
"transient": {
|
||||
"deleting": false
|
||||
}
|
||||
},
|
||||
"id": "8rwjrOD1ChjD"
|
||||
},
|
||||
"source": [
|
||||
"## RandomBatchGeoSampler"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"metadata": {
|
||||
"collapsed": true,
|
||||
"jupyter": {
|
||||
"source_hidden": false,
|
||||
"outputs_hidden": false
|
||||
},
|
||||
"nteract": {
|
||||
"transient": {
|
||||
"deleting": false
|
||||
}
|
||||
},
|
||||
"gather": {
|
||||
"logged": 1629249843438
|
||||
},
|
||||
"id": "v-N2fo6UChjE",
|
||||
"outputId": "497f6869-1ab7-4db7-bbce-e943b493ca41"
|
||||
},
|
||||
"source": [
|
||||
"for cache in [False, True]:\n",
|
||||
" chesapeake = ChesapeakeDE(os.path.join(ROOT, \"chesapeake\", \"DE\"), cache=cache)\n",
|
||||
" naip = NAIP(os.path.join(ROOT, \"naip\"), crs=chesapeake.crs, res=chesapeake.res, cache=cache)\n",
|
||||
" dataset = chesapeake + naip\n",
|
||||
" sampler = RandomBatchGeoSampler(naip.index, size=1000, batch_size=12, length=888)\n",
|
||||
" dataloader = DataLoader(dataset, batch_sampler=sampler)\n",
|
||||
" duration, count = time_epoch(dataloader)\n",
|
||||
" print(duration, count)"
|
||||
],
|
||||
"execution_count": null,
|
||||
"outputs": [
|
||||
{
|
||||
"output_type": "stream",
|
||||
"name": "stdout",
|
||||
"text": [
|
||||
"230.51380324363708 74\n",
|
||||
"53.99923872947693 74\n"
|
||||
]
|
||||
}
|
||||
]
|
||||
}
|
||||
]
|
||||
}
|
||||
Загрузка…
Ссылка в новой задаче