deep-learning-for-image-processing

deep-learning-for-image-processing copied to clipboard

**System information** * Have I written custom code: * OS Platform(e.g., window10 or Linux Ubuntu 16.04): * Python version: * Deep learning framework and version(e.g., Tensorflow2.1 or Pytorch1.3): * Use...

njugbs

yolov3_spp正负样本匹配device设置

2

> https://github.com/WZMIAOMIAO/deep-learning-for-image-processing/blob/80681e81327320a37cfc35b754d8204a5156e9c2/pytorch_object_detection/yolov3_spp/build_utils/utils.py#L287 在跑模型时跳的bug，`at`的device为cpu，下面需要添加一行 `at = at.to(targets.device)`

jiaerfei

fasterRCNN代码运行报错RuntimeError: unexpected EOF, expected 2193276 more bytes. The file might be corrupted.

1

*代码报错详情* ``` Traceback (most recent call last): File "D:/PythonProjects/deep-learning-for-image-processing-master/pytorch_object_detection/faster_rcnn/train_res50_fpn.py", line 231, in main(args) File "D:/PythonProjects/deep-learning-for-image-processing-master/pytorch_object_detection/faster_rcnn/train_res50_fpn.py", line 106, in main model = create_model(num_classes=args.num_classes + 1) File "D:/PythonProjects/deep-learning-for-image-processing-master/pytorch_object_detection/faster_rcnn/train_res50_fpn.py", line 30, in create_model...

theRain0450

您好，我使用unet在DRIVE数据集内容不变，resize为512*512的大小时运行train.py报错，交叉熵函数Target 244 is out of bounds.请问是不是激活函数设置成了与图像大小相关。如果想修改的话，在哪里可以修改

wangwxr

哥哥们知道是怎么回事吗

pastrami06

高版本的torch报错未发现torch._six

3

**System information** * Have I writtenTraceback (most recent call last): File "C:\Work\Pycharm\faster_rcnn\train_mobilenetv2.py", line 11, in from train_utils import GroupedBatchSampler, create_aspect_ratio_groups File "C:\Work\Pycharm\faster_rcnn\train_utils\__init__.py", line 4, in from .coco_eval import CocoEvaluator File...

water107

**问题1：为什么obj-loss一直抖动的很厉害呢**   **问题2：Gram-CAM对v3spp自定义数据集的训练模型进行热力图绘制，发现热力图没有在目标处高激活** 请问是网络没训练好的问题嘛？还是数据迁移的不够的呢？虽然mAP0.5可以到93% - 还有就是：3个yolo层，只有一个yolo层由特征，大尺度和小尺度层的特征提取都不尽人意；一方面是正样本就没在这两层的anchor上，anchor正负样本匹配的时候，就没几个，自然bbox loss和class loss也就没有了，只有objloss还有。 - 为什么yolo系列负样本只有obj-loss参与计算？ 

HaronCHou

iou

1

有没有大佬知道怎么改IOU损失呢，如果我不想使用标准的IOU计算想采用ciou，siou等，除了修改box.py中的box_iou函数。这个函数返回的是CIOU,SIOU。还需要对那些地方修改哦

thestars-maker

retinanet模型训练后进行验证,验证指标得到的结果全是0

4

训练retinanet模型时,每个epoch结束后在验证集上验证总会出现下边的这种问题 Average Precision (AP) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0 Average Precision (AP) @[ IoU=0.50 | area= all | maxDets=100 ] = 0 Average Precision (AP)...

161399695

TensorFlow 2.4.1使用mobilenetv3,自己训练的权重.ckpt转.h5，遇到错误

1

使用train_mobilenet_v3.py，进行的转换。网盘提供的权重可以转换，迁移学习的权重，如下错误： Traceback (most recent call last): File "/home/xt/tf2.3-master/mobilenet_v3/trans_v3_weights.py", line 66, in main() File "/home/xt/tf2.3-master/mobilenet_v3/trans_v3_weights.py", line 60, in main weights = rename_var(ckpt_path, m_info) File "/home/xt/tf2.3-master/mobilenet_v3/trans_v3_weights.py", line 35, in rename_var assert k...

QLW-07