deeplab predict a black mask

Question

i trained deeplab model on my custom dataset and does not predict anything just a black background, i don't know what's the problem

-data = RGB image + ( 0-1) label: 400 * 300

-classe=2

-convert to record format:

-training step:

==> Loss = 0.2 ～ 0.1

python train.py \ --logtostderr \ --vis_split = "train" \ --model_variant = "xception_65" \ --atrous_rates = 6 \ --atrous rates = 12 \ --atrous rates = 18 \ --output_stride = 16 \ --decoder_output_stride = 4 \ --training_number_of_steps = 1000 --train_crop_size = 513 \ --train_batch_size = 1 \ --train_crop_size = 513 \ --Fine_tune_batch_norm=False \ --Tf_initial_checkpoint = "./ Data / Init_models / Deelabv3_pascal_train_aug \ model.ckpt" --Initialize_last_layer = False \ --Last_layers_contain_logits_only = True \ --train_logdir="./data/log/train" \ --dataset_dir="./data/tfrecord" \ --dataset="pascal_voc_seg"

-convert to .pb step

python export_model.py \ --logtostderr \ -model_variant = "xception_65" \ --atrous_rates = 6 \ --atrous_rates = 12 \ --atrous_rates = 18 \ --output_stride = 16

until this step, everything looks fine

as output this what i got this

as settings screenshots

as settings screenshots

as settings screenshots

Answer 1

Convert your label to grayscale [0 - 255] and your two classes should have pixel values 0 and 1. Set num of classes=2 and ignore label=255. This worked for me.