Paddleocr Training - AssertionError: The None.pdparams does not exists!

[brookscar]

🔎 Search before asking

• I have searched the PaddleOCR Docs and found no similar bug report.
• I have searched the PaddleOCR Issues and found no similar bug report.
• I have searched the PaddleOCR Discussions and found no similar bug report.

🐛 Bug (问题描述)

Trying to train a model based off of a custom image set, have set up dataset > eval , train folders and used path in config. I keep getting the error:

Traceback (most recent call last):
File "/home/OCRServer/OCR/PaddleOCR/tools/train.py", line 269, in
main(config, device, logger, vdl_writer, seed)
File "/home/OCRServer/OCR/PaddleOCR/tools/train.py", line 215, in main
pre_best_model_dict = load_model(
File "/home/OCRServer/OCR/PaddleOCR/ppocr/utils/save_load.py", line 95, in load_model
assert os.path.exists(
AssertionError: The None.pdparams does not exists!

I also tried using the pretrained en_PP-OCRv3_det_slim_distill_train best_recall.pdparams as a pretrained model to get around the issue and it still did not work.

Config file below:
Global:
debug: true
use_gpu: false # Change to true if using GPU
is_float16: true
epoch_num: 200
log_smooth_window: 20
print_batch_step: 10
save_model_dir: ./output/det_ppocr_v4
save_epoch_step: 10
eval_batch_step: [0, 2000]
cal_metric_during_train: true
pretrained_model: '/home/OCRServer/en_PP-OCRv3_det_slim_distill_train/best_recall.pdparams' # Path to pretrained model
checkpoints: None
save_inference_dir: None
use_visualdl: false
infer_img: doc/imgs_words/ch/word_1.jpg
distributed: true
save_res_path: ./output/det/predicts_ppocrv3.txt

Optimizer:
name: Adam
beta1: 0.9
beta2: 0.999
lr:
name: Cosine
learning_rate: 0.001
warmup_epoch: 5
regularizer:
name: L2
factor: 3.0e-05

Architecture:
model_type: det # Text detection model type
algorithm: DB # Use 'DB' for DBNet (common for detection tasks)
Backbone:
name: MobileNetV3
Neck:
name: DBFPN
out_channels: 256
Head:
name: DBHead
out_channels: 2
k: 50

Loss:
name: DBLoss # Change to DBLoss for DBNet
balance_loss: true
main_loss_type: DiceLoss
alpha: 5.0
beta: 10.0
ohem_ratio: 3

PostProcess:
name: DBPostProcess
box_thresh: 0.6
unclip_ratio: 1.5
max_candidates: 1000

Metric:
name: DetMetric
main_indicator: hmean

Train:
dataset:
name: SimpleDataSet
data_dir: /home/OCRServer/OCR/dataset/train/test_images
label_file_list:
- /home/OCRServer/OCR/dataset/train_list.txt
transforms:
- DecodeImage:
img_mode: BGR
channel_first: false
- DetLabelEncode: {}
- DetResizeForTest:
image_shape: [640, 640] # Resize image for detection
- KeepKeys:
keep_keys:
- image
- shape
- polygons
- ignore_tags
loader:
shuffle: true
batch_size_per_card: 16 # Adjust based on available resources
drop_last: true
num_workers: 8

Eval:
dataset:
name: SimpleDataSet
data_dir: /home/OCRServer/OCR/dataset/eval/images
label_file_list:
- /home/OCRServer/OCR/dataset/eval_list.txt
transforms:
- DecodeImage:
img_mode: BGR
channel_first: false
- DetLabelEncode: {}
- DetResizeForTest:
image_shape: [640, 640]
- KeepKeys:
keep_keys:
- image
- shape
- polygons
- ignore_tags
loader:
shuffle: false
drop_last: false
batch_size_per_card: 16
num_workers: 4

🏃‍♂️ Environment (运行环境)

Ubuntu 22.04.5 LTS, Python 3.10.12, paddleocr 2.9.1, paddlepaddle 2.6.2

🌰 Minimal Reproducible Example (最小可复现问题的Demo)

python PaddleOCR/tools/train.py -c dataset/flightrec_config.yml

[GreatV]

The error might be caused by an incorrect checkpoints configuration. Since you have set:

checkpoints: None

Make sure this is correct. If you are resuming training from a checkpoint, this should point to the .pdparams file of the checkpoint. Otherwise, keep it as null.

[brookscar]

The error might be caused by an incorrect checkpoints configuration. Since you have set:

checkpoints: None

Make sure this is correct. If you are resuming training from a checkpoint, this should point to the .pdparams file of the checkpoint. Otherwise, keep it as null.

I have been trying to train the model off of custom image set and did not use a pretrained model or use checkpoints, so I set them to 'None'. This seems to be the issue, but do not know why I would get the Assertion error when I have set both parameters 'None'.