OCRv4文字检测微调后推理模型和训模型检测框不一致

[1061302569]

请提供下述完整信息以便快速定位问题/Please provide the following information to quickly locate the problem

• 系统环境/System Environment：linux
• 版本号/Version：Paddle：2.4 PaddleOCR：2.7 问题相关组件/Related components：
• 运行指令/Command Code：
• 完整报错/Complete Error Message：
  

不知道如何调参让他们能一致

目前尝试的方法一（失败）：
注释这两个参数

目前尝试的方法二（失败）
训练模型的配置没有image_shape加了这个也统一不了参数

下面使用的是v4检测模型配置，调了一下学习率其他都是默认参数，求指教推理模型和训练模型不一致，搜了所有教程也无法调好
{
"Global": {
"debug": false,
"use_gpu": true,
"epoch_num": 500,
"log_smooth_window": 20,
"print_batch_step": 100,
"save_model_dir": "output",
"save_epoch_step": 1,
"eval_batch_step": [
0,
1500
],
"cal_metric_during_train": false,
"checkpoints": null,
"pretrained_model": "https://paddleocr.bj.bcebos.com/pretrained/PPHGNet_small_ocr_det.pdparams",
"save_inference_dir": null,
"use_visualdl": false,
"distributed": true
},
"Architecture": {
"model_type": "det",
"algorithm": "DB",
"Transform": null,
"Backbone": {
"name": "PPHGNet_small",
"det": true
},
"Neck": {
"name": "LKPAN",
"out_channels": 256,
"intracl": true
},
"Head": {
"name": "PFHeadLocal",
"k": 50,
"mode": "large"
}
},
"Loss": {
"name": "DBLoss",
"balance_loss": true,
"main_loss_type": "DiceLoss",
"alpha": 5,
"beta": 10,
"ohem_ratio": 3
},
"Optimizer": {
"name": "Adam",
"beta1": 0.9,
"beta2": 0.999,
"lr": {
"name": "Cosine",
"learning_rate": 0.001,
"warmup_epoch": 2
},
"regularizer": {
"name": "L2",
"factor": "1e-6"
}
},
"PostProcess": {
"name": "DBPostProcess",
"thresh": 0.3,
"box_thresh": 0.6,
"max_candidates": 1000,
"unclip_ratio": 1.5
},
"Metric": {
"name": "DetMetric",
"main_indicator": "hmean"
},
"Train": {
"dataset": {
"name": "TextDetDataset",
"data_dir": "datasets/ICDAR2015",
"label_file_list": [
"datasets/ICDAR2015/train.txt"
],
"transforms": [
{
"DecodeImage": {
"img_mode": "BGR",
"channel_first": false
}
},
{
"DetLabelEncode": null
},
{
"CopyPaste": null
},
{
"IaaAugment": {
"augmenter_args": [
{
"type": "Fliplr",
"args": {
"p": 0.5
}
},
{
"type": "Affine",
"args": {
"rotate": [
-10,
10
]
}
},
{
"type": "Resize",
"args": {
"size": [
0.5,
3
]
}
}
]
}
},
{
"EastRandomCropData": {
"size": [
640,
640
],
"max_tries": 50,
"keep_ratio": true
}
},
{
"MakeBorderMap": {
"shrink_ratio": 0.4,
"thresh_min": 0.3,
"thresh_max": 0.7,
"total_epoch": 500
}
},
{
"MakeShrinkMap": {
"shrink_ratio": 0.4,
"min_text_size": 8,
"total_epoch": 500
}
},
{
"NormalizeImage": {
"scale": "1./255.",
"mean": [
0.485,
0.456,
0.406
],
"std": [
0.229,
0.224,
0.225
],
"order": "hwc"
}
},
{
"ToCHWImage": null
},
{
"KeepKeys": {
"keep_keys": [
"image",
"threshold_map",
"threshold_mask",
"shrink_map",
"shrink_mask"
]
}
}
]
},
"loader": {
"shuffle": true,
"drop_last": false,
"batch_size_per_card": 8,
"num_workers": 3
}
},
"Eval": {
"dataset": {
"name": "TextDetDataset",
"data_dir": "datasets/ICDAR2015",
"label_file_list": [
"datasets/ICDAR2015/val.txt"
],
"transforms": [
{
"DecodeImage": {
"img_mode": "BGR",
"channel_first": false
}
},
{
"DetLabelEncode": null
},
{
"DetResizeForTest": null
},
{
"NormalizeImage": {
"scale": "1./255.",
"mean": [
0.485,
0.456,
0.406
],
"std": [
0.229,
0.224,
0.225
],
"order": "hwc"
}
},
{
"ToCHWImage": null
},
{
"KeepKeys": {
"keep_keys": [
"image",
"shape",
"polys",
"ignore_tags"
]
}
}
]
},
"loader": {
"shuffle": false,
"drop_last": false,
"batch_size_per_card": 1,
"num_workers": 0
}
},
"profiler_options": null
}

[lipeng1109]

我v3微调使用的是你的第一种方式，是可以的，断点跟踪一下，看看两个在数据预处理上有什么区别