remap_lut = np.zeros((maxkey + 100), dtype=np.int32) remap_lut[list(class_remap.keys())] = list(class_remap.values()) # print(remap_lut) # create evaluator ignore = [] for cl, ign in class_ignore.items(): if ign: x_cl = int(cl) ignore.append(x_cl) print("Ignoring xentropy class ", x_cl, " in IoU evaluation") # create evaluator if FLAGS.backend == "torch": from auxiliary.torch_ioueval import iouEval evaluator = iouEval(nr_classes, ignore) if FLAGS.backend == "numpy": from auxiliary.np_ioueval import iouEval evaluator = iouEval(nr_classes, ignore) else: print("Backend for evaluator should be one of ", str(backends)) quit() evaluator.reset() # get test set test_sequences = DATA["split"][FLAGS.split] # get label paths label_names = [] for sequence in test_sequences: sequence = '{0:02d}'.format(int(sequence))
# print(remap_lut) # create evaluator ignore = [] for cl, ign in class_ignore.items(): if ign: x_cl = int(cl) ignore.append(x_cl) print("Ignoring xentropy class ", x_cl, " in IoU evaluation") # create evaluator evaluators = [] for i in range(len(DISTANCES)): if FLAGS.backend == "torch": from auxiliary.torch_ioueval import iouEval evaluators.append(iouEval(nr_classes, ignore)) evaluators[i].reset() elif FLAGS.backend == "numpy": from auxiliary.np_ioueval import iouEval evaluators.append(iouEval(nr_classes, ignore)) evaluators[i].reset() else: print("Backend for evaluator should be one of ", str(backends)) quit() # get test set test_sequences = DATA["split"][FLAGS.split] # get scan paths scan_names = [] for sequence in test_sequences: