How to train PASCAL VOC 2017 in Centernet2

[yangwenshuangshuang]

I try to train PASCAL VOC 2017 in Centernet2, and the the training process is normal. However, I use PascalVOCDetectionEvaluator and meet some problems during the test phase.

Here is the code:
def do_test(cfg, model):
results = OrderedDict()
for dataset_name in cfg.DATASETS.TEST:
mapper = None if cfg.INPUT.TEST_INPUT_TYPE == 'default' else
DatasetMapper(
cfg, False, augmentations=build_custom_augmentation(cfg, False))
# print(dataset_name)
data_loader = build_detection_test_loader(cfg, dataset_name, mapper=mapper)
output_folder = os.path.join(
cfg.OUTPUT_DIR, "inference_{}".format(dataset_name))
evaluator_type = MetadataCatalog.get(dataset_name).evaluator_type
# print(evaluator_type)
if evaluator_type == "lvis":
evaluator = LVISEvaluator(dataset_name, cfg, True, output_folder)
elif evaluator_type == 'coco':
evaluator = COCOEvaluator(dataset_name, cfg, True, output_folder)
elif evaluator_type == 'pascal_voc':
evaluator = PascalVOCDetectionEvaluator(dataset_name)
else:
assert 0, evaluator_type

    results[dataset_name] = inference_on_dataset(
        model, data_loader, evaluator)
    if comm.is_main_process():
        logger.info("Evaluation results for {} in csv format:".format(
            dataset_name))
        print_csv_format(results[dataset_name])
if len(results) == 1:
    results = list(results.values())[0]
return results

Here is the issue:
Traceback (most recent call last):
File "/media/ntu/volume1/home/s122md306_02/yangwenshuang/anaconda/anaconda3/envs/centernet2_new2/lib/python3.8/site-packages/torch/multiprocessing/spawn.py", line 59, in _wrap
fn(i, *args)
File "/media/ntu/volume1/home/s122md306_02/yangwenshuang/detectron2/detectron2/engine/launch.py", line 123, in _distributed_worker
main_func(*args)
File "/media/ntu/volume1/home/s122md306_02/yangwenshuang/detectron2/projects/CenterNet2/train_net.py", line 234, in main
return do_test(cfg, model)
File "/media/ntu/volume1/home/s122md306_02/yangwenshuang/detectron2/projects/CenterNet2/train_net.py", line 70, in do_test
results[dataset_name] = inference_on_dataset(
File "/media/ntu/volume1/home/s122md306_02/yangwenshuang/detectron2/detectron2/evaluation/evaluator.py", line 204, in inference_on_dataset
results = evaluator.evaluate()
File "/media/ntu/volume1/home/s122md306_02/yangwenshuang/detectron2/detectron2/evaluation/pascal_voc_evaluation.py", line 102, in evaluate
rec, prec, ap = voc_eval(
File "/media/ntu/volume1/home/s122md306_02/yangwenshuang/detectron2/detectron2/evaluation/pascal_voc_evaluation.py", line 220, in voc_eval
recs[imagename] = parse_rec(annopath.format(imagename))
File "/media/ntu/volume1/home/s122md306_02/yangwenshuang/detectron2/detectron2/evaluation/pascal_voc_evaluation.py", line 135, in parse_rec
tree = ET.parse(f)
OSError: [Errno 9] Bad file descriptor

Why there is some problem with PascalVOCDetectionEvaluator?

[bmaxdk]

Your error appeared that file handling issue. Can you check that your file path to the annotations in the annopath.format(imagename) is correct? You can also add debug print statements to see what annopath.format(imagename) actually is and check if the file exists at that path.