def test_fvd_demo1(run):
# Carla
test_obj = TestSTARSDataProc()
cfg = get_config(run)
if cfg["eval"]:
# Get p,r for one side
# test_1(run)
# Get trajectory for one side
# Get mota,motp for side 0 using GT
test_2(run)
# Get trajectory for other side
from StarsDataProcessing.detection_tracking.detection_tracking import (
detection_tracking,
)
sys.path.append("../StarsDataProcessing/detection_tracking")
cfg["intersection_config"] = cfg["intersection_config_2"]
cfg["video"] = cfg["video_2"]
cfg["trajectory_pred"] = cfg["trajectory_pred_2"]
detection_tracking(cfg)
assert check_file_and_size(cfg["trajectory_pred"])
# combine both
cfg2 = get_config(run)
traj_combined = combine(cfg2["trajectory_pred"], cfg2["trajectory_pred_2"])
traj_combined.to_csv(cfg2["trajectory_pred_combined"], index=False)
assert check_file_and_size(cfg2["trajectory_pred_combined"])
else:
# get trajectories
test_obj.test_get_trajectory(run)
# Get TL status for both
test_obj.test_get_TL_status(run)
# compute TLV, LVD using both sides
test_obj.test_get_TLV(run)
test_obj.test_get_LVD(run)
if cfg["eval"]:
# Compare params with GT
test_obj.test_validate_params_wrt_gt(run)
assert True
def test_4(run):
import pytest
import json
import os
import pandas as pd
import sys
from test_simulation import (
get_config,
check_file_and_size,
get_config_file,
get_change,
)
sys.path.append("../")
# Run Carla in the background
# Read behavior config
cfg = get_config(run)
behavior_cfg_path = cfg["behavior_config"]
behavior_cfg = get_config_file(behavior_cfg_path)
carla_cfg_path = cfg["carla_config"]
carla_cfg = get_config_file(carla_cfg_path)
print(carla_cfg)
config_data = {}
config_data.update(carla_cfg)
config_data.update(behavior_cfg)
from CarlaDataCollection.SimulationSubsystem.carla_params_verification import (
CarlaParamsVerifiction,
)
# Pass config to data collection functions, save the image frames
save_frames = False
observe_gt = True
carla_params_verification = CarlaParamsVerifiction(
config_data, observe_gt, save_frames
)
obs_paramTLJ, obs_paramLVD = carla_params_verification.run()
gt_paramLVD = behavior_cfg["global_distance_to_leading_vehicle"]
gt_paramTLJ = behavior_cfg["ignore_lights_percentage"]
LVD_change = get_change(gt_paramLVD, obs_paramLVD)
TLJ_change = get_change(gt_paramTLJ, obs_paramTLJ)
LVD_tol = 100
TLJ_tol = 20
print("LVD change : ", LVD_change, "%")
print("TLJ change : ", TLJ_change, "%")
assert LVD_change <= LVD_tol or TLJ_change <= TLJ_tol
def test_temp_2(run):
import json
from test_stars_data_proc import get_config, check_file_and_size, get_config_file
import sys
sys.path.append("../")
# Running Detection Tracking on video
from StarsDataProcessing.detection_tracking.detection_tracking_copy import (
detection_tracking,
)
cfg = get_config(run)
detection_tracking(cfg)
def test_6(run):
import pytest
import json
import os
import pandas as pd
import sys
from test_simulation import (
get_config,
check_file_and_size,
get_config_file,
get_change,
)
sys.path.append("../")
# Run Carla in the background
# Read behavior config
cfg = get_config(run)
behavior_cfg_path = cfg["behavior_config"]
behavior_cfg = get_config_file(behavior_cfg_path)
carla_cfg_path = cfg["carla_config"]
carla_cfg = get_config_file(carla_cfg_path)
print(carla_cfg)
config_data = {}
config_data.update(carla_cfg)
config_data.update(behavior_cfg)
from CarlaDataCollection.SimulationSubsystem.carla_params_verification import (
CarlaParamsVerifiction,
)
# Pass config to data collection functions, save the image frames
save_frames = False
observe_gt = False
carla_params_verification = CarlaParamsVerifiction(
config_data, observe_gt, save_frames
)
observed_fps = carla_params_verification.run()
required_fps = 10
print("FPS : ", observed_fps)
assert observed_fps > required_fps
def test_1(run):
from StarsDataProcessing.detection_modelling import visualize_evaluation
cfg = get_config(run)
checkpoints_path = "../StarsDataProcessing/detection_modelling/final_training/output/model_final.pth"
dataset_folder = cfg["detection_gt"]
p_vehicle, r_vehicle = visualize_evaluation.run(
dataset_folder,
checkpoints_path,
confidence_thres=0.8,
iou_thres=0.5,
show_visualization=True,
)
print(f"Final Precision: {p_vehicle}; Final Recall: {r_vehicle}")
assert p_vehicle > 0.75
assert r_vehicle > 0.75
def test_2(run):
import json
from test_stars_data_proc import get_config, check_file_and_size, get_config_file
import sys
sys.path.append("../")
# Running Detection Tracking on video
from StarsDataProcessing.detection_tracking.detection_tracking import (
detection_tracking,
)
cfg = get_config(run)
mota, motp = detection_tracking(cfg)
intersection_cfg_path = cfg["intersection_config"]
intersection_cfg = get_config_file(intersection_cfg_path)
intersection_type = intersection_cfg["type"]
if intersection_type == "carla":
assert mota >= 0.4
assert motp >= 0.4
assert True
def test_3(run):
import json
from test_stars_data_proc import get_config, check_file_and_size, get_config_file
import sys
sys.path.append("../")
# Running Detection Tracking on video
from StarsDataProcessing.detection_tracking.detection_tracking import (
detection_tracking,
)
cfg = get_config(run)
detection_tracking(cfg)
# Extracting params from trajectory files
from StarsDataProcessing.tl_violation.traffic_violation import tlv
from StarsDataProcessing import extract_params
tlv = tlv(cfg)
lvd_stats = extract_params.get_global_stats()
if check_file_and_size(cfg["params_pred"]):
with open(cfg["params_pred"]) as f:
params = json.load(f)
else:
params = {}
params["LVD"] = lvd_stats
params["TLV"] = tlv
os.remove(cfg["params_pred"])
with open(cfg["params_pred"], "w") as f:
json.dump(params, f, indent=4)
intersection_cfg_path = cfg["intersection_config"]
intersection_cfg = get_config_file(intersection_cfg_path)
intersection_type = intersection_cfg["type"]
if intersection_type == "carla":
params_pred = get_config_file(cfg["params_pred"])
params_gt = get_config_file(cfg["params_gt"])
dist_param = "min_lvd_global"
if params_gt["TLV"] != 0:
eps = 0
else:
eps = 0.001
diff_tlv = (params_pred["TLV"] - params_gt["TLV"] - eps) / (
params_gt["TLV"] + eps
)
if params_gt["LVD"] != 0:
eps = 0
else:
eps = 0.001
diff_lvd = (params_pred["LVD"][dist_param] / 9.46 - params_gt["LVD"] - eps) / (
params_gt["LVD"] + eps
)
print("TLV : ", diff_tlv, "LVD : ", diff_lvd)
assert abs(diff_tlv) <= 0.2
assert abs(diff_lvd) <= 0.2