def do_kitti_detection_evaluation(dataset, predictions, output_folder, logger):
predict_folder = os.path.join(output_folder, 'data') # only recognize data
mkdir(predict_folder)
for image_id, prediction in predictions.items():
predict_txt = image_id + '.txt'
predict_txt = os.path.join(predict_folder, predict_txt)
generate_kitti_3d_detection(prediction, predict_txt)
logger.info("Evaluate on KITTI dataset")
output_dir = os.path.abspath(output_folder)
os.chdir('./smoke/data/datasets/evaluation/kitti/kitti_eval')
label_dir = "/home/niels/Documents/Uni/MA/Implementation/SMOKE/datasets/kitti/testing/label_2"
if not os.path.isfile('evaluate_object_3d_offline'):
subprocess.Popen(
'g++ -O3 -DNDEBUG -o evaluate_object_3d_offline evaluate_object_3d_offline.cpp',
shell=True)
command = "./evaluate_object_3d_offline {} {}".format(
label_dir, output_dir)
output = subprocess.check_output(command,
shell=True,
universal_newlines=True).strip()
logger.info(output)
os.chdir('../../../../../../tools')
def kitti_evaluation(dataset, predictions, output_dir):
"""Do evaluation by process kitti eval program
Args:
dataset (paddle.io.Dataset): [description]
predictions (Paddle.Tensor): [description]
output_dir (str): path of save prediction
"""
# Clear data dir before do evaluate
if os.path.exists(os.path.join(output_dir, 'data')):
shutil.rmtree(os.path.join(output_dir, 'data'))
predict_folder = os.path.join(output_dir, 'data') # only recognize data
mkdir(predict_folder)
type_id_conversion = getattr(dataset, 'TYPE_ID_CONVERSION')
id_type_conversion = {value:key for key, value in type_id_conversion.items()}
for image_id, prediction in predictions.items():
predict_txt = image_id + '.txt'
predict_txt = os.path.join(predict_folder, predict_txt)
generate_kitti_3d_detection(prediction, predict_txt, id_type_conversion)
output_dir = os.path.abspath(output_dir)
root_dir = os.getcwd()
os.chdir('./tools/kitti_eval_offline')
label_dir = getattr(dataset, 'label_dir')
label_dir = os.path.join(root_dir, label_dir)
if not os.path.isfile('evaluate_object_3d_offline'):
subprocess.Popen('g++ -O3 -DNDEBUG -o evaluate_object_3d_offline evaluate_object_3d_offline.cpp', shell=True)
command = "./evaluate_object_3d_offline {} {}".format(label_dir, output_dir)
os.system(command)
def do_kitti_detection_evaluation(dataset,
predictions,
output_folder,
logger
):
predict_folder = os.path.join(output_folder, 'data') # only recognize data
mkdir(predict_folder)
for image_id, prediction in predictions.items():
predict_txt = image_id + '.txt'
predict_txt = os.path.join(predict_folder, predict_txt)
generate_kitti_3d_detection(prediction, predict_txt)
logger.info("Evaluate on KITTI dataset")
output_dir = os.path.abspath(output_folder)
original_dir = os.getcwd()
os.chdir('./smoke/data/datasets/evaluation/kitti/kitti_eval')
label_dir = getattr(dataset, 'label_dir')
if not os.path.isfile('evaluate_object_offline'):
subprocess.Popen('g++ -O3 -DNDEBUG -o evaluate_object_offline evaluate_object_offline.cpp', shell=True)
print(os.listdir())
#command = "./evaluate_object_offline {} {}".format("/app/datasets/kitti/training/label_2", output_dir)
#output = subprocess.check_output(command, shell=True, universal_newlines=True).strip()
#logger.info(output)
os.chdir(original_dir)
def default_setup(cfg, args):
output_dir = cfg.OUTPUT_DIR
if output_dir:
mkdir(output_dir)
rank = comm.get_rank()
logger = setup_logger(output_dir, rank)
logger.info("Using {} GPUs".format(args.num_gpus))
logger.info("Collecting environment info")
logger.info("\n" + collect_env_info())
logger.info(args)
logger.info("Loaded configuration file {}".format(args.config_file))
with open(args.config_file, "r") as cf:
config_str = "\n" + cf.read()
logger.info(config_str)
logger.info("Running with config:\n{}".format(cfg))
# make sure each worker has a different, yet deterministic seed if specified
seed_all_rng(None if cfg.SEED < 0 else cfg.SEED + rank)
# cudnn benchmark has large overhead. It shouldn't be used considering the small size of
# typical validation set.
if not (hasattr(args, "eval_only") and args.eval_only):
torch.backends.cudnn.benchmark = cfg.CUDNN_BENCHMARK
def do_kitti_detection_evaluation(dataset, predictions, output_folder, logger):
predict_folder = os.path.join(output_folder, 'data') # only recognize data
mkdir(predict_folder)
for image_id, prediction in predictions.items():
predict_txt = image_id + '.txt'
predict_txt = os.path.join(predict_folder, predict_txt)
generate_kitti_3d_detection(prediction, predict_txt)
logger.info("Evaluate on KITTI dataset")
output_dir = os.path.abspath(output_folder)
print("---ANI! output_dir - ", output_dir, "---")
print("---ANI! os.getcwd() - ", os.getcwd(), "---")
cur_dir = os.getcwd()
# ch_dir = os.path.join(cur_dir, "./smoke/data/datasets/evaluation/kitti/kitti_eval")
os.chdir('./smoke/data/datasets/evaluation/kitti/kitti_eval')
# os.chdir(ch_dir)
print("---ANI! output_dir after first change - ", output_dir, "---")
print("---ANI! os.getcwd() after first change - ", os.getcwd(), "---")
label_dir = getattr(dataset, 'label_dir')
logger.info(
"---ANI! label_dir before manual change is {} ---".format(label_dir))
# TODO Change name to evaluate_object_offline for 40 points
# TODO Change name to evaluate_object_3d_offline for 11 points
executable_name = "evaluate_object_3d_offline"
if not os.path.isfile(executable_name):
# subprocess.Popen('g++ -O3 -DNDEBUG -o evaluate_object_3d_offline evaluate_object_3d_offline.cpp', shell=True)
subprocess.call('g++ -O3 -DNDEBUG -o {} {}.cpp'.format(
executable_name, executable_name),
shell=True)
logger.info("Compiling executable for {} for first time!".format(
executable_name))
else:
logger.info(
"Compiled executable {} already exists!".format(executable_name))
logger.info(
"---ANI! label_dir: {} ---\n---ANI! output_dir: {} ---\n".format(
label_dir, output_dir))
label_dir = os.path.join(cur_dir, 'datasets/kitti/training/label_2')
label_dir = os.path.abspath(label_dir)
command = "./{} {} {}".format(executable_name, label_dir, output_dir)
logger.info("---ANI! command: {} ---".format(command))
output = subprocess.check_output(command,
shell=True,
universal_newlines=True).strip()
logger.info(output)
ch_dir = os.path.join(cur_dir, "tools")
# os.chdir('../tools')
os.chdir(ch_dir)
print("---ANI! output_dir after second change - ", output_dir, "---")
print("---ANI! os.getcwd() after second change - ", os.getcwd(), "---")
def do_kitti_detection_evaluation(dataset,
predictions,
output_folder,
logger,
eval=True):
predict_folder = os.path.join(output_folder, 'data') # only recognize data
mkdir(predict_folder)
for image_id, prediction in predictions.items():
predict_txt = image_id + '.txt'
predict_txt = os.path.join(predict_folder, predict_txt)
generate_kitti_3d_detection(prediction, predict_txt)
if eval:
# logger.info("Evaluate on KITTI dataset")
# output_dir = os.path.abspath(output_folder)
# os.chdir('../smoke/data/datasets/evaluation/kitti/kitti_eval')
# label_dir = getattr(dataset, 'label_dir')
# if not os.path.isfile('evaluate_object_3d_offline'):
# subprocess.Popen('g++ -O3 -DNDEBUG -o evaluate_object_3d_offline evaluate_object_3d_offline.cpp', shell=True)
# command = "./evaluate_object_3d_offline {} {}".format(label_dir, output_dir)
# output = subprocess.check_output(command, shell=True, universal_newlines=True).strip()
# logger.info(output)
# os.chdir('../tools')
logger.info("Evaluate on KITTI dataset")
output_dir = os.path.abspath(output_folder)
smoke_dir = os.getcwd()
print(os.getcwd())
os.chdir(
os.path.join(smoke_dir, 'smoke/data/datasets/evaluation/kitti/'))
label_dir = getattr(dataset, 'label_dir')
eval_results = evaluate(
os.path.join(smoke_dir, label_dir),
os.path.join(output_dir, 'data'),
label_split_file=os.path.join(
smoke_dir, 'datasets/kitti/training/ImageSets/val.txt'))
logger.info(eval_results)
os.chdir(smoke_dir)
return
def run_test(cfg, model):
eval_types = ("detection",)
output_folders = [None] * len(cfg.DATASETS.TEST)
dataset_names = cfg.DATASETS.TEST
if cfg.OUTPUT_DIR:
for idx, dataset_name in enumerate(dataset_names):
output_folder = os.path.join(cfg.OUTPUT_DIR, "inference", dataset_name)
mkdir(output_folder)
output_folders[idx] = output_folder
data_loaders_val = build_test_loader(cfg)
# import pdb; pdb.set_trace()
for output_folder, dataset_name, data_loader_val in zip(output_folders, dataset_names, data_loaders_val):
inference(
model,
data_loaders_val,
dataset_name=dataset_name,
eval_types=eval_types,
device=cfg.MODEL.DEVICE,
output_folder=output_folder,
)
comm.synchronize()