Maximum number of steps (fixations)

[ManooshSamiei]

Hello,

I am wondering why you increase the maximum eye trajectory length by one, changing it from 6 to 7, while you calculate the cdf, in the below code:

def compute_search_cdf(scanpaths, annos, max_step, return_by_task=False):
# compute search CDF
task_names = np.unique([traj['task'] for traj in scanpaths])
num_steps = get_num_steps(scanpaths, annos, task_names)
cdf_tasks = get_mean_cdf(num_steps, task_names, max_step + 1)
if return_by_task:
    return dict(zip(task_names, cdf_tasks))
else:
    mean_cdf = np.mean(cdf_tasks, axis=0)
    std_cdf = np.std(cdf_tasks, axis=0)
    return mean_cdf, std_cdf

its written

cdf_tasks = get_mean_cdf(num_steps, task_names, max_step + 1)

which adds 1 to the max_step.

[ouyangzhibo]

Hello,

I am wondering why you increase the maximum eye trajectory length by one, changing it from 6 to 7, while you calculate the cdf, in the below code:

def compute_search_cdf(scanpaths, annos, max_step, return_by_task=False):
# compute search CDF
task_names = np.unique([traj['task'] for traj in scanpaths])
num_steps = get_num_steps(scanpaths, annos, task_names)
cdf_tasks = get_mean_cdf(num_steps, task_names, max_step + 1)
if return_by_task:
    return dict(zip(task_names, cdf_tasks))
else:
    mean_cdf = np.mean(cdf_tasks, axis=0)
    std_cdf = np.std(cdf_tasks, axis=0)
    return mean_cdf, std_cdf

its written

cdf_tasks = get_mean_cdf(num_steps, task_names, max_step + 1)

which adds 1 to the max_step.

That't due to the fact that the initial fixations starts at the center and we wanted to see the results for 6 extra steps (excluding the initial one), so we add one step after the max_step in one of our experiments. You can remove the +1 here if it is not desired in your case.

[ManooshSamiei]

I see.. Thank you very much. I first thought when you mentioned you used 6 fixations in the paper, you are including the initial fixation as well.