Datasets:

ArXiv:
License:
zhengthomastang's picture
Include evaluation tools for 2024 edition
a3341ee
import os
import re
import argparse
import shutil
import json
import csv
import numpy as np
import pandas as pd
from typing import List, Set, Any
class ValidateFile(argparse.Action):
"""
Module to validate files
"""
def __call__(self, parser, namespace, values, option_string = None):
if not os.path.exists(values):
parser.error(f"Please enter a valid file path. Got: {values}")
elif not os.access(values, os.R_OK):
parser.error(f"File {values} doesn't have read access")
setattr(namespace, self.dest, values)
def validate_file_path(input_string: str) -> str:
"""
Validates whether the input string matches a file path pattern
:param str input_string: input string
:return: validated file path
:rtype: str
::
file_path = validate_file_path(input_string)
"""
file_path_pattern = r"^[a-zA-Z0-9_\-\/.#]+$"
if re.match(file_path_pattern, input_string):
return input_string
else:
raise ValueError(f"Invalid file path: {input_string}")
def load_csv_to_dataframe_from_file(file_path: str, column_names: List[str], camera_ids: Set, interval: int = 1) -> pd.DataFrame:
"""
Loads dataframe from a CSV file
:param str file_path: file path
:param List[str] column_names: column names
:return: dataframe in the file
:rtype: pd.DataFrame
::
dataFrame = load_csv_to_dataframe_from_file(file_path, column_names)
"""
data: List[List[str]] = list()
valid_file_path = validate_file_path(file_path)
df = pd.read_csv(valid_file_path, sep=" ", header=None, names=column_names, dtype={"CameraId": int, "Id": int, "FrameId": int})
# Ensure non-negative values for CameraId, Id, FrameId
if (df[['CameraId', 'Id', 'FrameId']] < 0).any().any():
raise ValueError("Invalid negative values found for CameraId, Id, or FrameId.")
# Filter by camera_id
df = df[df['CameraId'].isin(camera_ids)]
# Filter rows where FrameId % interval == 0
df = df[df['FrameId'] % interval == 0]
# Round the last two columns (assuming these are 'Xworld' and 'Yworld')
df['Xworld'] = df['Xworld'].round(3)
df['Yworld'] = df['Yworld'].round(3)
if len(df) == 0:
raise ValueError("DataFrame is empty after filtering process.")
return df
def write_dataframe_to_csv_to_file(file_path: str, data: pd.DataFrame, delimiter: str = " ") -> None:
"""
Writes dataframe to a CSV file
:param str file_path: file path
:param pd.DataFrame data: dataframe to be written
:param str delimiter: delimiter of the CSV file
:return: None
::
write_dataframe_to_csv_to_file(file_path, data, delimiter)
"""
data.to_csv(file_path, sep=delimiter, index=False, header=False)
def make_dir(dir_path: str) -> None:
"""
Makes a directory without removing other files
:param str dir_path: directory path
:return: None
::
make_dir(dir_path)
"""
valid_dir_path = validate_file_path(dir_path)
if not os.path.isdir(valid_dir_path):
os.makedirs(validate_file_path(dir_path))
def make_seq_maps_file(file_dir: str, scenes: List[str], benchmark: str, split_to_eval: str) -> None:
"""
Makes a sequence-maps file used by TrackEval library
:param str file_dir: file path
:param Set(str) sensor_ids: names of sensors
:param str benchmark: name of the benchmark
:param str split_to_eval: name of the split of data
:return: None
::
make_seq_maps_file(file_dir, sensor_ids)
"""
make_clean_dir(file_dir)
file_name = benchmark + "-" +split_to_eval + ".txt"
seq_maps_file = file_dir + "/" + file_name
f = open(seq_maps_file, "w")
f.write("name\n")
for name in scenes:
sensor_name = str(name) + "\n"
f.write(sensor_name)
# f.write("FINAL")
f.close()
def make_seq_ini_file(gt_dir: str, scene: str, seq_length: int) -> None:
"""
Makes a sequence-ini file used by TrackEval library
:param str gt_dir: file path
:param str scene: Name of a single scene
:param int seq_length: Number of frames
:return: None
::
make_seq_ini_file(gt_dir, scene, seq_length)
"""
ini_file_name = gt_dir + "/seqinfo.ini"
f = open(ini_file_name, "w")
f.write("[Sequence]\n")
name= "name=" +str(scene)+ "\n"
f.write(name)
f.write("imDir=img1\n")
f.write("frameRate=30\n")
seq = "seqLength=" + str(seq_length) + "\n"
f.write(seq)
f.write("imWidth=1920\n")
f.write("imHeight=1080\n")
f.write("imExt=.jpg\n")
f.close()
def get_scene_to_camera_id_dict(file_path):
"""
Loads a mapping of scene names to camera IDs from a JSON file.
:param str file_path: Path to the JSON file containing scenes data.
:return: A dictionary where keys are scene names and values are lists of camera IDs.
::
scene_to_camera_id_dict = get_scene_to_camera_id_dict(file_path)
"""
scene_2_cam_id = dict()
valid_file_path = validate_file_path(file_path)
with open(valid_file_path, "r") as file:
scenes_data = json.load(file)
for scene_data in scenes_data:
scene_name = scene_data["scene_name"]
camera_ids = scene_data["camera_ids"]
if scene_name not in scene_2_cam_id:
scene_2_cam_id[scene_name] = []
scene_2_cam_id[scene_name].extend(camera_ids)
return scene_2_cam_id
def check_file_size(file_path):
"""
Checks the size of a file and raises an exception if it exceeds 2 GB.
:param str file_path: Path to the file to be checked.
:return: None
:raises ValueError: If the file size is greater than 2 GB.
::
check_file_size(file_path)
"""
file_size_bytes = os.path.getsize(file_path)
file_size_gb = file_size_bytes / (2**30)
if file_size_gb > 2:
raise ValueError(f"The size of the file is {file_size_gb:.2f} GB, which is greater than the 2 GB")
def make_clean_dir(dir_path: str) -> None:
"""
Makes a clean directory
:param str dir_path: directory path
:return: None
::
make_clean_dir(dir_path)
"""
valid_dir_path = validate_file_path(dir_path)
if os.path.exists(valid_dir_path):
shutil.rmtree(dir_path, ignore_errors=True)
if not os.path.isdir(valid_dir_path):
os.makedirs(validate_file_path(dir_path))