administrator/frigate
1
0
Fork 0
mirror of https://github.com/blakeblackshear/frigate.git synced 2026-05-02 03:27:41 +03:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

Better handle model dir structure

Browse Source
This commit is contained in:
Nicolas Mowen 2025-05-29 10:51:40 -06:00
parent 53b8b1335f
commit 60a58ad2ab
1 changed files with 12 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

19
frigate/util/classification.py
View File

@ -15,9 +15,9 @@ LEARNING_RATE = 0.001
@staticmethod @staticmethod
def generate_representative_dataset(train_dir: str): def generate_representative_dataset(dataset_dir: str):
image_paths = [] image_paths = []
for root, dirs, files in os.walk("train"): for root, dirs, files in os.walk(dataset_dir):
for file in files: for file in files:
if file.lower().endswith((".jpg", ".jpeg", ".png")): if file.lower().endswith((".jpg", ".jpeg", ".png")):
image_paths.append(os.path.join(root, file)) image_paths.append(os.path.join(root, file))
@ -32,10 +32,15 @@ def generate_representative_dataset(train_dir: str):
@staticmethod @staticmethod
def train_classification_model(train_dir: str) -> bool: def train_classification_model(model_dir: str) -> bool:
"""Train a classification model.""" """Train a classification model."""
dataset_dir = os.path.join(model_dir, "dataset")
num_classes = len( num_classes = len(
[d for d in os.listdir(train_dir) if os.path.isdir(os.path.join(train_dir, d))] [
d
for d in os.listdir(dataset_dir)
if os.path.isdir(os.path.join(dataset_dir, d))
]
) )
# Start with imagenet base model with 35% of channels in each layer # Start with imagenet base model with 35% of channels in each layer
@ -77,7 +82,7 @@ def train_classification_model(train_dir: str) -> bool:
class_indices = train_gen.class_indices class_indices = train_gen.class_indices
index_to_class = {v: k for k, v in class_indices.items()} index_to_class = {v: k for k, v in class_indices.items()}
sorted_classes = [index_to_class[i] for i in range(len(index_to_class))] sorted_classes = [index_to_class[i] for i in range(len(index_to_class))]
with open(os.path.join(train_dir, "labelmap.txt"), "w") as f: with open(os.path.join(model_dir, "labelmap.txt"), "w") as f:
for class_name in sorted_classes: for class_name in sorted_classes:
f.write(f"{class_name}\n") f.write(f"{class_name}\n")
@ -87,12 +92,12 @@ def train_classification_model(train_dir: str) -> bool:
# convert model to tflite # convert model to tflite
converter = tf.lite.TFLiteConverter.from_keras_model(model) converter = tf.lite.TFLiteConverter.from_keras_model(model)
converter.optimizations = [tf.lite.Optimize.DEFAULT] converter.optimizations = [tf.lite.Optimize.DEFAULT]
converter.representative_dataset = generate_representative_dataset converter.representative_dataset = generate_representative_dataset(dataset_dir)
converter.target_spec.supported_ops = [tf.lite.OpsSet.TFLITE_BUILTINS_INT8] converter.target_spec.supported_ops = [tf.lite.OpsSet.TFLITE_BUILTINS_INT8]
converter.inference_input_type = tf.uint8 converter.inference_input_type = tf.uint8
converter.inference_output_type = tf.uint8 converter.inference_output_type = tf.uint8
tflite_model = converter.convert() tflite_model = converter.convert()
# write model # write model
with open(os.path.join(train_dir, "model.tflite"), "wb") as f: with open(os.path.join(model_dir, "model.tflite"), "wb") as f:
f.write(tflite_model) f.write(tflite_model)