{
"cells": [
{
"cell_type": "markdown",
"id": "ffac96cc",
"metadata": {},
"source": [
"# 2D Single Person Human Pose Estimation\n",
"\n",
"Megh Shukla: work.meghshukla@gmail.com"
]
},
{
"cell_type": "code",
"execution_count": 1,
"id": "4f1a0174",
"metadata": {
"scrolled": true
},
"outputs": [],
"source": [
"# Ignore syntax warning related to HRNet syntax\n",
"import warnings\n",
"warnings.filterwarnings('ignore')\n",
"\n",
"# Python imports\n",
"import os\n",
"import cv2\n",
"import copy\n",
"import logging\n",
"\n",
"# External package imports\n",
"import numpy as np\n",
"import pandas as pd\n",
"from tqdm import tqdm\n",
"import matplotlib\n",
"from matplotlib import pyplot as plt\n",
"\n",
"# PyTorch imports\n",
"import torch\n",
"import torch.utils.data\n",
"from torch.utils.tensorboard import SummaryWriter\n",
"from torch.optim.lr_scheduler import ReduceLROnPlateau as ReduceLROnPlateau\n",
"\n",
"# Imports from supporting files\n",
"from config import ParseConfig\n",
"\n",
"from dataloader import load_hp_dataset\n",
"from dataloader import HumanPoseDataLoader\n",
"\n",
"from activelearning import ActiveLearning\n",
"\n",
"from evaluation import PercentageCorrectKeypoint\n",
"\n",
"from utils import fast_argmax\n",
"from utils import visualize_image\n",
"from utils import heatmap_loss\n",
"from utils import count_parameters\n",
"from utils import get_pairwise_joint_distances\n",
"\n",
"from train_test import Train\n",
"from train_test import Metric\n",
"from train_test import load_models\n",
"from train_test import define_hyperparams\n",
"\n",
"\n",
"from models.auxiliary.AuxiliaryNet import AuxNet\n",
"from models.hrnet.pose_hrnet import PoseHighResolutionNet as HRNet\n",
"from models.stacked_hourglass.StackedHourglass import PoseNet as Hourglass\n",
"\n",
"\n",
"logging.getLogger().setLevel(logging.INFO)"
]
},
{
"cell_type": "markdown",
"id": "429418c5",
"metadata": {},
"source": [
"### Initializing configurations:\n",
"\n",
"Model: Hourglass / HRNet
\n",
"Dataset: MPII / Leeds Sports Pose
\n",
"Load and Save paths
\n",
"... and much more!\n",
"\n",
"Have a look at ```configuration.yml```\n"
]
},
{
"cell_type": "code",
"execution_count": 2,
"id": "b5118c76",
"metadata": {
"scrolled": true
},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"INFO:root:Saving the model at: D:\\WADLA Demo\\Demo_3\n"
]
}
],
"source": [
"conf = ParseConfig()"
]
},
{
"cell_type": "markdown",
"id": "452f21ae",
"metadata": {},
"source": [
"### Load datasets:\n",
"\n",
"Based on our choice of dataset in ```configuration.yml```, we will load our selection into memory.
\n",
"Quick recall, different datasets have different data formats!
\n",
"```load_hp_dataset``` is located in ```dataloader.py```"
]
},
{
"cell_type": "code",
"execution_count": 3,
"id": "51215d87",
"metadata": {
"scrolled": true
},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"INFO:root:Loading MPII dataset\n",
"INFO:root:Loading precached MPII.\n"
]
}
],
"source": [
"dataset_dict = load_hp_dataset(dataset_conf=conf.dataset, model_conf=conf.model)"
]
},
{
"cell_type": "markdown",
"id": "a020c918",
"metadata": {},
"source": [
"### Initialize (...and load) model architecture:\n",
"\n",
"We initialize and load the model based on the choice of architecture specified in ```configuration.yml```.
\n",
"```load_models``` is located in ```train_test.py```"
]
},
{
"cell_type": "code",
"execution_count": 4,
"id": "c57205cb",
"metadata": {
"scrolled": true
},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"INFO:root:Initializing Hourglass Network\n",
"INFO:root:Loading Pose model from: D:\\WADLA Demo\\Pretrained_MPII_256channels\\\n",
"INFO:root:Successfully loaded Pose model.\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Number of parameters (Hourglass): 8429088\n",
"\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"INFO:root:Successful: Model transferred to GPUs.\n",
"\n"
]
}
],
"source": [
"pose_model, aux_net = load_models(conf=conf, load_pose=conf.model['load'], load_aux=conf.model['aux_net']['load'],\n",
" model_dir=conf.model['load_path'])"
]
},
{
"cell_type": "markdown",
"id": "2002899b",
"metadata": {},
"source": [
"### Define the active learning library:\n",
"\n",
"Although we won't be doing active learning, the library will allow us to draw samples randomly for training the model.
\n",
"The class ```ActiveLearning``` is located in ```activelearning.py```."
]
},
{
"cell_type": "code",
"execution_count": 5,
"id": "b46aad9f",
"metadata": {
"scrolled": true
},
"outputs": [],
"source": [
"activelearning = ActiveLearning(conf=conf, pose_net=pose_model, aux_net=aux_net)"
]
},
{
"cell_type": "markdown",
"id": "ec12dc82",
"metadata": {},
"source": [
"### Creating an object that inherits from torch.utils.data.Dataset()\n",
"\n",
"Writing code in PyTorch is fairly simple, and PyTorch requires two objects that need to be explicitly coded: _model_ and _dataset_.
\n",
"For our scenario, we create an object of type ```HumanPoseDataLoader``` for handling our dataset.
\n",
"\n",
"```datasets``` standardizes ```mpii``` and ```lsp``` into a common format, defines augmentation routines, calls ```activelearning``` to sample from this data, and also controls preprocessing at batch level.\n",
"\n",
"```HumanPoseDataLoader``` is located in ```dataloader.py```"
]
},
{
"cell_type": "code",
"execution_count": 6,
"id": "3a23dcc0",
"metadata": {
"scrolled": true
},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"INFO:root:Creating MPII dataset\n",
"\n",
"INFO:root:MPII dataset description:\n",
"INFO:root:Length (#images): 15247\n",
"INFO:root:Creating single person patches\n",
"\n",
"INFO:root:Selecting train and validation images where all joints are present.\n",
"\n",
"INFO:root:Size of MPII processed dataset: \n",
"INFO:root:Train: 10614\n",
"INFO:root:Validate: 2416\n",
"\n",
"INFO:root:Creating train and validation splits\n",
"\n",
"INFO:root:Initializing base dataset.\n",
"INFO:root:\n",
"Final size of Training Data: 1000\n",
"INFO:root:Final size of Validation Data: 2416\n",
"\n",
"INFO:root:\n",
"Dataloader Initialized.\n",
"\n"
]
}
],
"source": [
"datasets = HumanPoseDataLoader(dataset_dict=dataset_dict, activelearning=activelearning, conf=conf)"
]
},
{
"cell_type": "markdown",
"id": "6885e2f5",
"metadata": {},
"source": [
"Once we initialize our dataloader, we delete a few objects that we won't need any longer to clear memory.
\n",
"Deleting the models gets rid of any computational graphs computed during ```ActiveLearning``` which we don't require any longer."
]
},
{
"cell_type": "code",
"execution_count": 7,
"id": "5fec9505",
"metadata": {
"scrolled": false
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"