{
"cells": [
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"#Libs\n",
"import cv2\n",
"import numpy as np\n",
"import pandas as pd\n",
"from tabulate import tabulate\n",
"import matplotlib.pyplot as plt"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Methods"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"def split_on_nan_row(recording):\n",
" # Split the DataFrame into subdatasets\n",
" subdatasets = []\n",
" current_subdataset = []\n",
"\n",
" for index, row in recording.iterrows():\n",
" if pd.isna(row['x1']) and pd.isna(row['y1']) and pd.isna(row['x2']) and pd.isna(row['y2']):\n",
" if current_subdataset:\n",
" subdatasets.append(pd.DataFrame(current_subdataset))\n",
" current_subdataset = []\n",
" elif pd.isna(row['x1']) and pd.isna(row['y1']) and row['x2'] and row['y2']:\n",
" if current_subdataset:\n",
" subdatasets.append(pd.DataFrame(current_subdataset))\n",
" current_subdataset = []\n",
" elif row['x1'] and row['y1'] and pd.isna(row['x2']) and pd.isna(row['y2']):\n",
" if current_subdataset:\n",
" subdatasets.append(pd.DataFrame(current_subdataset))\n",
" current_subdataset = []\n",
" else:\n",
" if index > 0:\n",
" prev_row = recording.iloc[index - 1]\n",
" if not (pd.isna(prev_row['x1']) and pd.isna(prev_row['y1']) and pd.isna(prev_row['x2']) and pd.isna(prev_row['y2'])):\n",
" current_subdataset.append(row)\n",
"\n",
" # Append the last subdataset if it exists\n",
" if current_subdataset:\n",
" subdatasets.append(pd.DataFrame(current_subdataset))\n",
"\n",
" for i in range(len(subdatasets)):\n",
" if len(subdatasets[i]) < 10:\n",
" subdatasets.pop(i)\n",
" break\n",
" return subdatasets"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"def get_average_points(subdatasets):\n",
" average_points = []\n",
"\n",
" for i in subdatasets:\n",
" x1 = i['x1'].mean() \n",
" y1 = i['y1'].mean()\n",
" x2 = i['x2'].mean()\n",
" y2 = i['y2'].mean()\n",
" average_points.append((x1, y1, x2, y2))\n",
"\n",
" return average_points"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"def correct_points_orientation(list_of_points, label):\n",
" average_points_corrected = []\n",
" for i in range(len(list_of_points)):\n",
" x1, y1, x2, y2 = list_of_points[i]\n",
" if x1 < x2:\n",
" x1, x2 = x2, x1\n",
" y1, y2 = y2, y1\n",
" average_points_corrected.append((x1, y1, x2, y2, label[i]))\n",
" return average_points_corrected"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Variables"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# Angular error\n",
"angle_error = np.radians(0.25) # 0.25°\n",
"\n",
"# Define the positions of the eyes (world coordinates)\n",
"coord_left_eye = np.array([-3.25, 0, 0])\n",
"coord_right_eye = np.array([3.25, 0, 0])"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Recordings"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"label1 = ['infini', '25c', '25up', '25r', '50c', '50up', '50r', '75c', '75up', '75r', '200c', '200up', '200r' ]\n",
"\n",
"recording1_left = pd.read_csv('./recording1/centroids_left.csv')\n",
"recording1_right = pd.read_csv('./recording1/centroids_right.csv')\n",
"\n",
"# Retenir seulement les images non noirs\n",
"subdatasets1_left = split_on_nan_row(recording1_left)\n",
"subdatasets1_right = split_on_nan_row(recording1_right)\n",
"\n",
"# Recupérer les points moyens pour les douze points de fixations\n",
"average_points_left1 = get_average_points(subdatasets1_left)\n",
"average_points_right1 = get_average_points(subdatasets1_right)\n",
"\n",
"# Verifier qu'il n'y a pas d'échanges entre les centroides 1 et 2\n",
"average_points_left_corrected1, average_points_right_corrected1 = correct_points_orientation(average_points_left1, label1), correct_points_orientation(average_points_right1, label1)\n",
"\n",
"xy_eye_left1 = np.column_stack(([(average_points_left_corrected1[i][0] + average_points_left_corrected1[i][2])/2 for i in range(len(average_points_left_corrected1))], [(average_points_left_corrected1[i][1] + average_points_left_corrected1[i][3])/2 for i in range(len(average_points_left_corrected1))]))\n",
"xy_eye_right1 = np.column_stack(([(average_points_right_corrected1[i][0] + average_points_right_corrected1[i][2])/2 for i in range(len(average_points_right_corrected1))], [(average_points_right_corrected1[i][1] + average_points_right_corrected1[i][3])/2 for i in range(len(average_points_right_corrected1))]))"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Recording 2"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"label2 = ['infini', '25c', '50c', '75c', '200c', '25up', '50up', '75up', '200up', '25r', '50r', '75r', '200r' ]\n",
"\n",
"recording2_left = pd.read_csv('./recording2/centroids_left.csv')\n",
"recording2_left = recording2_left.drop([11474])\n",
"recording2_left = recording2_left.reset_index(drop=True)\n",
"\n",
"recording2_right = pd.read_csv('./.recording2/centroids_right.csv')\n",
"recording2_right = recording2_right.drop([319, 5926, 7611, 10201, 11500, 12791])\n",
"recording2_right = recording2_right.reset_index(drop=True)\n",
"\n",
"subdatasets2_left = split_on_nan_row(recording2_left)\n",
"subdatasets2_right = split_on_nan_row(recording2_right)\n",
"\n",
"average_points_left2 = get_average_points(subdatasets2_left)\n",
"average_points_right2 = get_average_points(subdatasets2_right)\n",
"\n",
"average_points_left_corrected2, average_points_right_corrected2 = correct_points_orientation(average_points_left2, label2), correct_points_orientation(average_points_right2, label2)\n",
"\n",
"xy_eye_left2 = np.column_stack(([(average_points_left_corrected2[i][0] + average_points_left_corrected2[i][2])/2 for i in range(len(average_points_left_corrected2))], [(average_points_left_corrected2[i][1] + average_points_left_corrected2[i][3])/2 for i in range(len(average_points_left_corrected2))]))\n",
"xy_eye_right2 = np.column_stack(([(average_points_right_corrected2[i][0] + average_points_right_corrected2[i][2])/2 for i in range(len(average_points_right_corrected2))], [(average_points_right_corrected2[i][1] + average_points_right_corrected2[i][3])/2 for i in range(len(average_points_right_corrected2))]))"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Recording 3"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"label3 = ['infini', '75c', '75up', '50up', '75r', '50r', '200c', '25c', '25r', '50c', '200r', '25up', '200up']\n",
"\n",
"recording3_left = pd.read_csv('./recording3/centroids_left.csv')\n",
"recording3_right = pd.read_csv('./recording3/centroids_right.csv')\n",
"recording3_right = recording3_right.drop([9268, 10181, 11001, 12687])\n",
"recording3_right = recording3_right.reset_index(drop=True)\n",
"\n",
"subdatasets3_left = split_on_nan_row(recording3_left)\n",
"subdatasets3_right = split_on_nan_row(recording3_right)\n",
"\n",
"average_points_left3 = get_average_points(subdatasets3_left)\n",
"average_points_right3 = get_average_points(subdatasets3_right)\n",
"\n",
"average_points_left_corrected3, average_points_right_corrected3 = correct_points_orientation(average_points_left3, label3), correct_points_orientation(average_points_right3, label3)\n",
"\n",
"xy_eye_left3 = np.column_stack(([(average_points_left_corrected3[i][0] + average_points_left_corrected3[i][2])/2 for i in range(len(average_points_left_corrected3))], [(average_points_left_corrected3[i][1] + average_points_left_corrected3[i][3])/2 for i in range(len(average_points_left_corrected3))]))\n",
"xy_eye_right3 = np.column_stack(([(average_points_right_corrected3[i][0] + average_points_right_corrected3[i][2])/2 for i in range(len(average_points_right_corrected3))], [(average_points_right_corrected3[i][1] + average_points_right_corrected3[i][3])/2 for i in range(len(average_points_right_corrected3))]))"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Recording 4"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"label4 = ['infini', '25up', '200c', '200up', '75c', '25c', '50c', '200r', '25r', '75r', '75up', '50up', '50r' ]\n",
"\n",
"recording4_left = pd.read_csv('./recording4/centroids_left.csv')\n",
"recording4_left = recording4_left.drop([1971,1972, 4402, 6891,7982,9011, 10229, 10434, 12390, 13182])\n",
"recording4_left = recording4_left.reset_index(drop=True)\n",
"\n",
"recording4_right = pd.read_csv('./recording4/centroids_right.csv')\n",
"recording4_right = recording4_right.drop([1960, 4394, 8170, 10423, 13177])\n",
"recording4_right = recording4_right.reset_index(drop=True)\n",
"\n",
"subdatasets4_left = split_on_nan_row(recording4_left)\n",
"subdatasets4_right = split_on_nan_row(recording4_right)\n",
"\n",
"average_points_left4 = get_average_points(subdatasets4_left)\n",
"average_points_right4 = get_average_points(subdatasets4_right)\n",
"\n",
"average_points_left_corrected4, average_points_right_corrected4 = correct_points_orientation(average_points_left4, label4), correct_points_orientation(average_points_right4, label4)\n",
"\n",
"xy_eye_left4 = np.column_stack(([(average_points_left_corrected4[i][0] + average_points_left_corrected4[i][2])/2 for i in range(len(average_points_left_corrected4))], [(average_points_left_corrected4[i][1] + average_points_left_corrected4[i][3])/2 for i in range(len(average_points_left_corrected4))]))\n",
"xy_eye_right4 = np.column_stack(([(average_points_right_corrected4[i][0] + average_points_right_corrected4[i][2])/2 for i in range(len(average_points_right_corrected4))], [(average_points_right_corrected4[i][1] + average_points_right_corrected4[i][3])/2 for i in range(len(average_points_right_corrected4))]))"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "img_pro",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.12.0"
}
},
"nbformat": 4,
"nbformat_minor": 2
}