Release/0.3.0

[baudcode]

Major Features

• Documentation
• New experimental models: Unet++, PSP, FCN, U2Net
• Tensorflow 2.4 support

Models

• Unet++
• PSP
• FCN
• U2Net

Datasets

• BioimageBenchmark

Fixes

• Renamed Plastic ds to Taco ds
• Fixed train.py errors
• Fixed mixed precision training

Features

• Add learning rate finder
• Add optimal batch size finder
• Improved dataset loading
• Add patch as resize method
• Add colab notebook and

[codecov[bot]]

Codecov Report

Merging #6 (480d886) into master (f31d3e3) will decrease coverage by 3.50%. The diff coverage is 34.43%.

@@            Coverage Diff             @@
##           master       #6      +/-   ##
==========================================
- Coverage   51.45%   47.94%   -3.51%     
==========================================
  Files          84       94      +10     
  Lines        4674     5723    +1049     
==========================================
+ Hits         2405     2744     +339     
- Misses       2269     2979     +710     

Impacted Files	Coverage Δ
tf_semantic_segmentation/bin/tfrecord_writer.py	58.33% <0.00%> (ø)
tf_semantic_segmentation/bin/train_all.py	0.00% <0.00%> (ø)
tf_semantic_segmentation/datasets/taco.py	36.66% <ø> (ø)
...f_semantic_segmentation/debug/preprocessing_vis.py	0.00% <0.00%> (ø)
tf_semantic_segmentation/evaluation/eval_loss.py	0.00% <0.00%> (ø)
tf_semantic_segmentation/models/deeplabv3.py	0.00% <0.00%> (ø)
tf_semantic_segmentation/models/deeplabv3plus.py	0.00% <0.00%> (ø)
tf_semantic_segmentation/serving.py	17.64% <ø> (ø)
tf_semantic_segmentation/callbacks.py	22.27% <14.78%> (-11.88%)	:arrow_down:
tf_semantic_segmentation/models/apps/resnet50.py	79.36% <15.38%> (-6.73%)	:arrow_down:
... and 29 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 06b3480...480d886. Read the comment docs.