HongChow

puzzled me ...

TNN的mat和opencv的mat如何互转呢？

2

好像没看到太多的example？

请问如何保存tnn mat为图像？

2

如题

OS: UBUNTU18.04 Scanning dependencies of target openpose_lib [ 12%] Creating directories for 'openpose_lib' [ 25%] No download step for 'openpose_lib' [ 37%] No patch step for 'openpose_lib' [ 50%] No...

如何一次性生成超过20张图片?

1

默认是20

[save the entire model failed]

2

load and inference work ok. and I am trying to save the entire model (including both architecture and weights),but failed. in detail,torch.save(net,'mb_v1_ssd_all.pth') when run the code, it failed , the...

更低的loss不能更好的检测?

1

有两类:一类是图表, 一类是公式.数据集中公式比图表多一些. (1)数据集的处理:数据集很少,只有几百张图(train503张,测试150左右) 原始图片大小各异, 我先做了补白(取长边补),保证是数据集是长宽相等的图片,再做了标注. (2) 训练的命令: %run train_ssd.py --dataset_type voc --datasets ./data --net "vgg16-ssd" --base_net models/vgg16_reducedfc.pth --batch_size 6 --freeze_base_net --num_epochs 300 --lr 0.001 (用了vgg16_reducedfc.pth的与训练系数, basenet freeze了其他层的系数参与更新) 其他修改: (1) image_size:...

RuntimeError: Error(s) in loading state_dict for SSD: my command(using in spyder terminal): %run train_ssd.py --dataset_type voc --datasets ./data --net vgg16-ssd --pretrained_ssd models/vgg16_reducedfc.pth --batch_size 24 --freeze_net --num_epochs 200 --lr 0.001 it...

DeepLab seems crashed?

3

I am trying to finetune my own dataset(zm) to by the pretrained resenet model, but it returned some error information as following: Starting Epoch: 0 Total Epoches: 50 0%| |...

我记得有个专门针对100K做数据增强的网站,还有人记得么?