det 检测模型微调偶发异常：list index out of range

[kerry-weic]

🔎 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)

错误信息如下图：

list index out of range 错误出现14次，看错误信息好像是数据问题，如上图其中一条数据如下：

[2024/12/14 11:36:49] ppocr ERROR: When parsing line train/fe9ed4f7921749e09d76202cf9a00285_1.jpg	[{"transcription": "正常支付租金。", "points": [[306.0, 510.0], [917.0, 548.0], [910.0, 663.0], [299.0, 625.0]]}, {"transcription": "期开业阶段的租金、收银系统使用费、推广服务费及其它各项费用仍须自装修期届", "points": [[287.0, 1039.0], [3484.0, 1108.0], [3480.0, 1248.0], [283.0, 1179.0]]}, {"transcription": "满后的次日起照常缴纳。", "points": [[310.0, 1232.0], [1257.0, 1268.0], [1253.0, 1374.0], [305.0, 1337.0]]}, {"transcription": "4.租金", "points": [[299.0, 1410.0], [649.0, 1424.0], [645.0, 1535.0], [294.0, 1521.0]]}, {"transcription": "4.1.双方同意按“基础租金+提成租金”的模式计算租金。", "points": [[302.0, 1616.0], [2501.0, 1656.0], [2499.0, 1762.0], [300.0, 1721.0]]}, {"transcription": "4.2.基础租金部分", "points": [[303.0, 1788.0], [1035.0, 1806.0], [1032.0, 1912.0], [300.0, 1893.0]]}, {"transcription": "4.2.1.每月基础租金为：人民币（大写）柒拾陆万伍仟肆佰拾贰元", "points": [[302.0, 1966.0], [2983.0, 1995.0], [2982.0, 2100.0], [301.0, 2072.0]]}, {"transcription": "（￥765432元）。其他费用￥100元", "points": [[307.0, 2138.0], [1762.0, 2150.0], [1761.0, 2256.0], [306.0, 2244.0]]}, {"transcription": "4.2.2.付款方式：", "points": [[302.0, 2311.0], [1000.0, 2317.0], [999.0, 2422.0], [301.0, 2416.0]]}, {"transcription": "按月支付：每月提前至少10天支付下月基础租金。", "points": [[308.0, 2477.0], [2264.0, 2477.0], [2264.0, 2599.0], [308.0, 2599.0]]}, {"transcription": "4.3.提成租金部分", "points": [[318.0, 2657.0], [1044.0, 2657.0], [1044.0, 2761.0], [318.0, 2761.0]]}, {"transcription": "4.3.1.计算方式：租金=承租方收入*提成比例。", "points": [[307.0, 2833.0], [2113.0, 2822.0], [2114.0, 2927.0], [307.0, 2939.0]]}, {"transcription": "4. 3. 2.提成比例：16‱（）。", "points": [[307.0, 2999.0], [1536.0, 2993.0], [1537.0, 3105.0], [307.0, 3111.0]]}, {"transcription": "4.3.3.承租方收入", "points": [[312.0, 3178.0], [1050.0, 3166.0], [1052.0, 3271.0], [314.0, 3284.0]]}, {"transcription": "承租方收入是指：结算周期内的全部销售收入，不扣除其他成本；", "points": [[317.0, 3335.0], [2805.0, 3306.0], [2807.0, 3424.0], [318.0, 3452.0]]}, {"transcription": "承租方开设于其他地址的店铺收入不予计算。", "points": [[323.0, 3504.0], [2027.0, 3475.0], [2029.0, 3594.0], [325.0, 3623.0]]}, {"transcription": "4.3.4.结算周期：1个月。", "points": [[317.0, 3677.0], [1322.0, 3654.0], [1325.0, 3770.0], [320.0, 3794.0]]}, {"transcription": "5.租金支付", "points": [[316.0, 4025.0], [864.0, 3999.0], [870.0, 4120.0], [322.0, 4146.0]]}, {"transcription": "5.1.首期费用", "points": [[322.0, 4217.0], [877.0, 4192.0], [882.0, 4309.0], [327.0, 4335.0]]}, {"transcription": "首期费用对应租赁期限：年月日至年月日", "points": [[328.0, 4379.0], [2179.0, 4315.0], [2183.0, 4432.0], [332.0, 4496.0]]}, {"transcription": "5.2.之后费用支付", "points": [[322.0, 4735.0], [1041.0, 4692.0], [1048.0, 4798.0], [328.0, 4841.0]]}, {"transcription": "3.3.1.装修期仅作为乙方完成装修的期限要求;装修期作为租赁期限的一部分仍需", "points": [[291.0, 328.0], [3525.0, 404.0], [3525.0, 569.0], [299.0, 486.0]]}, {"transcription": "3.3.2. 乙方必须于装修期届满之日前完成装修工程并开业。乙方如有合理特殊情况", "points": [[302.0, 700.0], [3494.0, 758.0], [3490.0, 905.0], [302.0, 843.0]]}, {"transcription": "不能如期开业,须提前至少7日书面告知甲方且须获甲方书面同意延期开业,但延", "points": [[293.0, 866.0], [3507.0, 937.0], [3503.0, 1071.0], [293.0, 1025.0]]}, {"transcription": "4.3.5.结算付款方式:每个结算周期结束后10个工作日内,双方结算支付租金。", "points": [[332.0, 3815.0], [3428.0, 3815.0], [3431.0, 3953.0], [332.0, 3962.0]]}, {"transcription": "首期费用：共计人民币￥152元，", "points": [[327.0, 4559.0], [1616.0, 4501.0], [1607.0, 4618.0], [339.0, 4672.0]]}, {"transcription": "包含上述期限内的租金与租赁配套服务费用", "points": [[1628.0, 4484.0], [3348.0, 4542.0], [3365.0, 4718.0], [1632.0, 4626.0]]}, {"transcription": "租金与配套服务费用应按月结算支付，", "points": [[331.0, 4889.0], [1766.0, 4810.0], [1779.0, 4931.0], [339.0, 5011.0]]}, {"transcription": "应于每月10日前结算并支付。", "points": [[1804.0, 4781.0], [2980.0, 4898.0], [2975.0, 5036.0], [1804.0, 4919.0]]}]

🏃‍♂️ Environment (运行环境)

Python 3.8.15
paddlepaddle-gpu 2.6.1.post117

paddleocr 版本main分支

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

python3 -m paddle.distributed.launch --gpus '0'  tools/train.py -c configs/det/ch_PP-OCRv4/ch_PP-OCRv4_det_teacher.yml -o Global.pretrained_model=./pretrained_model/ch_PP-OCRv4_det_server_train/best_accuracy

[GreatV]

能提供一个最小可复现的数据集，我们debug一下吗

[kerry-weic]

现在训练是在docker里面，我先拉到本地debug试试，单看上面报错的数据好像没发现问题

[jingsongliujing]

可以看一下是不是数据集格式的问题，与可以成功读取训练的数据比对一下，格式上的异同，由于没有可复现的数据集，我们没法进行排查