Skip to content

Update dataset to v2.1 #13

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
pynicolas merged 5 commits into from
Feb 25, 2026
Merged

Update dataset to v2.1 #13

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
d43437f
Use stable dependencies
pynicolas Feb 25, 2026
c0117d4
Update dataset to v2.1
pynicolas Feb 25, 2026
9f21801
test: 1 epoch
pynicolas Feb 25, 2026
08e8f53
Python 3.10 -> 3.12
pynicolas Feb 25, 2026
1b972d7
Restore NB_EPOCHS=35
pynicolas Feb 25, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion .github/workflows/train.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -27,7 +27,7 @@ jobs:
- name: Set up Python
uses: actions/setup-python@v4
with:
python-version: '3.10'
python-version: '3.12'

- name: Install dependencies
run: |
Expand Down
13 changes: 6 additions & 7 deletions requirements.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,7 +1,6 @@
ai-edge-torch-nightly
tf_nightly
torchvision
segmentation-models-pytorch
albumentations
opencv-python
numpy
litert-torch==0.8.0
torchvision==0.24.1
segmentation-models-pytorch==0.5.0
albumentations==2.0.8
opencv-python==4.13.0.92
numpy==2.4.2
15 changes: 8 additions & 7 deletions train.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -28,7 +28,7 @@
from urllib.request import urlretrieve
import shutil

DATASET_VERSION = 'v2.0'
DATASET_VERSION = 'v2.1'
DATASET_ZIP_URL = f'https://github.com/pynicolas/fairscan-dataset/releases/download/{DATASET_VERSION}/fairscan-dataset-{DATASET_VERSION}.zip'

BUILD_DIR = "build"
Expand All @@ -38,7 +38,7 @@
DATASET_ZIP_PATH = BUILD_DIR + "/dataset.zip"
DATASET_PARENT_DIR = BUILD_DIR + "/dataset"
DATASET_DIR = DATASET_PARENT_DIR + "/fairscan-dataset"
NB_EPOCHS = 25
NB_EPOCHS = 35

if os.path.isdir(BUILD_DIR):
shutil.rmtree(BUILD_DIR)
Expand Down Expand Up @@ -181,7 +181,8 @@ def evaluate_encoder(encoder_name, model_save_path, device=torch.device('cpu')):
end = time.time()

print(f"- Epoch {epoch + 1}/{NB_EPOCHS}: train_loss={avg_train_loss:.4f} | Val Loss: {val_loss:.4f}" +
f" | Dice (cont): {dice_cont_mean:.4f} | Dice (disc): {dice_disc_mean:.4f} | {end - start:.1f} seconds")
f" | Dice (cont): {dice_cont_mean:.4f} | Dice (disc): {dice_disc_mean:.4f} | {end - start:.1f} seconds",
flush=True)

if dice_disc_mean > best_dice:
best_dice = dice_disc_mean
Expand Down Expand Up @@ -232,8 +233,8 @@ def evaluate_encoder(encoder_name, model_save_path, device=torch.device('cpu')):

# Convert to TFLite

import ai_edge_torch
from ai_edge_torch.generative.quantize import quant_recipes
import litert_torch
from litert_torch.generative.quantize import quant_recipes

model = smp.DeepLabV3Plus(
encoder_name=encoder,
Expand Down Expand Up @@ -280,10 +281,10 @@ def representative_dataset():
yield (img,)

# 3. quant_config
quant_config = quant_recipes.full_int8_dynamic_recipe()
quant_config = quant_recipes.full_dynamic_recipe()

# 4. Conversion
edge_model_quantized = ai_edge_torch.convert(
edge_model_quantized = litert_torch.convert(
wrapped_model,
sample_args=sample_args,
sample_kwargs=None,
Expand Down