EyeTrackVR/EyeTrackApp/leap.py
2024-05-30 09:52:47 -05:00

277 lines
10 KiB
Python

"""
------------------------------------------------------------------------------------------------------
,@@@@@@
@@@@@@@@@@@ @@@
@@@@@@@@@@@@ @@@@@@@@@@@
@@@@@@@@@@@@@ @@@@@@@@@@@@@@
@@@@@@@/ ,@@@@@@@@@@@@@
/@@@@@@@@@@@@@@@ @@@@@@@@
@@@@@@@@@@@@@@@@@@@@@@@@ @@@@@
@@@@@@@@ @@@@@
,@@@ @@@@&
@@@@@@. @@@@
@@@ @@@@@@@@@/ @@@@@
,@@@. @@@@@@((@ @@@@(
//@@@ ,, @@@@ @@@@@
@@@( @@@@@@@
@@@ @ @@@@@@@@#
@@@@@@@@@@@@@@@@@
@@@@@@@@@@@@@(
LEAP by: Prohurtz
Algorithm App Implementation By: Prohurtz
Copyright (c) 2023 EyeTrackVR <3
LICENSE: GNU GPLv3
------------------------------------------------------------------------------------------------------
"""
# LEAP = Lightweight Eyelid And Pupil
import os
os.environ["OMP_NUM_THREADS"] = "1"
import onnxruntime
import numpy as np
import cv2
import time
import math
from queue import Queue
import threading
from one_euro_filter import OneEuroFilter
import psutil, os
import sys
from utils.misc_utils import resource_path
from pathlib import Path
frames = 0
models = Path("Models")
def run_model(input_queue, output_queue, session):
while True:
frame = input_queue.get()
if frame is None:
break
img_np = np.array(frame)
img_np = img_np.astype(np.float32) / 255.0
gray_img = 0.299 * img_np[:, :, 0] + 0.587 * img_np[:, :, 1] + 0.114 * img_np[:, :, 2]
# Add the channel and batch dimensions
gray_img = np.expand_dims(gray_img, axis=0) # Add channel dimension
img_np = np.expand_dims(gray_img, axis=0) # Add batch dimension
# img_np = np.transpose(img_np, (2, 0, 1))
# img_np = np.expand_dims(img_np, axis=0)
ort_inputs = {session.get_inputs()[0].name: img_np}
pre_landmark = session.run(None, ort_inputs)
# pre_landmark = pre_landmark[1]
# pre_landmark = np.reshape(pre_landmark, (12, 2))
pre_landmark = np.reshape(pre_landmark, (-1, 2))
output_queue.put((frame, pre_landmark))
class LEAP_C(object):
def __init__(self):
onnxruntime.disable_telemetry_events()
# Config variables
self.num_threads = 4 # Number of python threads to use (using ~1 more than needed to achieve wanted fps yields lower cpu usage)
self.queue_max_size = 1 # Optimize for best CPU usage, Memory, and Latency. A maxsize is needed to not create a potential memory leak.
self.model_path = resource_path(models / 'LEAP053024.onnx')
self.low_priority = (
False # set process priority to low (may cause issues when unfocusing? reported by one, not reproducable)
)
self.low_priority = (
True # set process priority to low (may cause issues when unfocusing? reported by one, not reproducable)
)
self.print_fps = False
# Init variables
self.frames = 0
self.queues = []
self.threads = []
self.model_output = np.zeros((12, 2))
self.output_queue = Queue(maxsize=self.queue_max_size)
self.start_time = time.time()
for _ in range(self.num_threads):
self.queue = Queue(maxsize=self.queue_max_size)
self.queues.append(self.queue)
opts = onnxruntime.SessionOptions()
opts.inter_op_num_threads = 1
opts.intra_op_num_threads = 1
opts.graph_optimization_level = onnxruntime.GraphOptimizationLevel.ORT_ENABLE_ALL
opts.optimized_model_filepath = ""
if self.low_priority:
try:
process = psutil.Process(os.getpid()) # set process priority to low
try:
sys.getwindowsversion()
except AttributeError:
process.nice(0) # UNIX: 0 low 10 high
process.nice()
else:
process.nice(psutil.BELOW_NORMAL_PRIORITY_CLASS) # Windows
process.nice()
except:
pass
# See https://learn.microsoft.com/en-us/windows/win32/api/processthreadsapi/nf-processthreadsapi-getpriorityclass#return-value for values
else:
pass
# process = psutil.Process(os.getpid()) # set process priority to low
# try:
# sys.getwindowsversion()
# except AttributeError:
# process.nice(10) # UNIX: 0 low 10 high
# else:
# process.nice(psutil.HIGH_PRIORITY_CLASS) # Windows
# See https://learn.microsoft.com/en-us/windows/win32/api/processthreadsapi/nf-processthreadsapi-getpriorityclass#return-value for values
min_cutoff = 0.1
beta = 15.0
self.one_euro_filter = OneEuroFilter(np.random.rand(12, 2), min_cutoff=min_cutoff, beta=beta)
self.one_euro_filter_float = OneEuroFilter(np.random.rand(1, 2), min_cutoff=5, beta=0.007)
self.dmax = 0
self.dmin = 0
self.openlist = []
self.x = 0
self.y = 0
self.maxlist = []
self.ort_session1 = onnxruntime.InferenceSession(self.model_path, opts, providers=["CPUExecutionProvider"])
threads = []
for i in range(self.num_threads):
thread = threading.Thread(
target=run_model,
args=(self.queues[i], self.output_queue, self.ort_session1),
name=f"Thread {i}",
)
threads.append(thread)
thread.start()
def to_numpy(self, tensor):
return tensor.detach().cpu().numpy() if tensor.requires_grad else tensor.cpu().numpy()
def run_onnx_model(self, queues, session, frame):
for i in range(len(queues)):
if not queues[i].full():
queues[i].put(frame)
break
def leap_run(self):
img = self.current_image_gray_clean.copy()
img = cv2.cvtColor(img, cv2.COLOR_GRAY2RGB)
img_height, img_width = img.shape[:2] # Move outside the loop
frame = cv2.resize(img, (112, 112))
imgvis = self.current_image_gray.copy()
self.run_onnx_model(self.queues, self.ort_session1, frame)
if not self.output_queue.empty():
frame, pre_landmark = self.output_queue.get()
# pre_landmark = np.reshape(pre_landmark, (-1, 2))
# pre_landmark = self.one_euro_filter(pre_landmark)
for point in pre_landmark:
x, y = (point*112).astype(int)
x, y = int(x), int(y) # Ensure x and y are integers
cv2.circle(imgvis, (int(x), int(y)), 2, (255, 255, 0), -1)
x1, y1 = pre_landmark[1]
x2, y2 = pre_landmark[3]
x3, y3 = pre_landmark[4]
x4, y4 = pre_landmark[2]
d1 = math.dist(pre_landmark[1], pre_landmark[3])
# a more fancy method could be used taking into acount the relative size of the landmarks so that weirdness can be acounted for better
d2 = math.dist(pre_landmark[2], pre_landmark[4])
d = (d1 + d2) / 2
# by averaging both sets we can get less error? i think part of why 1 eye was better than the other is because we only considered one offset points.
# considering both should smooth things out between eyes
try:
if d >= np.percentile(
self.openlist, 80 # do not go above 85, but this value can be tuned
): # an aditional approach could be using the place where on average it is most stable, denoting what distance is the most stable "open"
self.maxlist.append(d)
if len(self.maxlist) > 2000: # i feel that this is very cpu intensive. think of a better method
self.maxlist.pop(0)
# this should be the average most open value, the average of top 2000 values in rolling calibration
# with this we can use it as the "openstate" (0.7, for expanded squeeze)
# weighted values to shift slightly to max value
normal_open = ((sum(self.maxlist) / len(self.maxlist)) * 0.90 + max(self.openlist) * 0.10) / (
0.95 + 0.15
)
except:
normal_open = 0.8
if len(self.openlist) < 5000: # TODO expose as setting?
self.openlist.append(d)
else:
self.openlist.pop(0)
self.openlist.append(d)
try:
per = (d - normal_open) / (min(self.openlist) - normal_open)
oldper = (d - max(self.openlist)) / (
min(self.openlist) - max(self.openlist)
) # TODO: remove when testing is done
per = 1 - per
per = per - 0.2 # allow for eye widen? might require a more legit math way but this makes sense.
per = min(per, 1.0) # clamp to 1.0 max
per = max(per, 0.0) # clamp to 1.0 min
# print("new: ", per, "vs old: ", oldper)
except:
per = 0.8
pass
x = pre_landmark[6][0]
y = pre_landmark[6][1]
self.last_lid = per
calib_array = np.array([per, per]).reshape(1, 2)
per = self.one_euro_filter_float(calib_array)
per = per[0][0]
# print(per)
if per <= 0.2: # TODO: EXPOSE AS SETTING
per == 0.0
# this should be tuned, i could make this auto calib based on min from a list of per values.
return imgvis, float(x), float(y), per
imgvis = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)
return imgvis, 0, 0, 0
class External_Run_LEAP(object):
def __init__(self):
self.algo = LEAP_C()
def run(self, current_image_gray, current_image_gray_clean):
self.algo.current_image_gray = current_image_gray
self.algo.current_image_gray_clean = current_image_gray_clean
img, x, y, per = self.algo.leap_run()
return img, x, y, per