sunny
sunny
也就是说300-W训练集+MENPO一起训练的模型效果在测试平常人脸效果更好咯,那您针对MENOP的训练集,有像300-W的训练集做镜像、平移之类的操作来扩充训练集训练吗?
哦哦 好的好的,感谢感谢,我试试~
您好,能提供下Menpo的训练数据集嘛,网上的链接貌似失效了!谢谢了!
哦哦,那可能是因为未翻墙,谢谢!
直接训练S1和S2有什么区别嘛?就层数的不同,之前试了部分数据集S1有时候误差比S2小一些,目前选的是直接训练S2
@Hanlos hello, thanks for you sharing, but how can i change the yolov3_to_onnx.py file? can you provide guidance, thanks.
config.yaml 内容如下: AUTO_RESUME: true CUDNN: BENCHMARK: true DETERMINISTIC: false ENABLED: true DATA_DIR: '' GPUS: (0,1) OUTPUT_DIR: 'LiteHRNet_w18_output' LOG_DIR: 'LiteHRNet_w18_log' WORKERS: 8 PRINT_FREQ: 300 DATASET: COLOR_RGB: false DATASET: 'coco' ROOT: '/mnt/share/COCO/'...
> darknet-》caffe-》mnn 看git 主页readme有详细教程 好的 谢谢~ 您有尝试cpu上使用MNN推理加速,MobileNetV2-YOLOv3-Fastest处理时间多少ms?
Hi Marek: I mainly run un ImageDemo.py, I , I tested a picture, the timing is started by by model.loadNetwork, rk, the time is mainly spent on the he model.processcessImg...
@xingyizhou In other words,the locations (batch['reg_ind']) values and the regression depth(batch['reg_target']) have some relationships? from the paper: ConerNet?