wuji3 / visiondk Goto Github PK
View Code? Open in Web Editor NEWA powerful baseline for image classification and face recognition with Pytorch
License: GNU General Public License v3.0
A powerful baseline for image classification and face recognition with Pytorch
License: GNU General Public License v3.0
thanks for your time, I know the repo supports some models in torchvision style, but I want to use my own model, what shall I do?
"""
model:
choice: torchvision-swin_t # torchvision- or custom-
kwargs: {} # kwargs feed into torchvision-style models.init
num_classes: 5
pretrained: True
backbone_freeze: False
bn_freeze: False
bn_freeze_affine: False
attention_pool: False
data:
root: data
nw: 2
imgsz:
- 480
- 480
train:
bs: 16 # one gpu if DDP
common_aug: null
class_aug: null
augment: # refer to utils/transforms.py
random_choice:
- random_color_jitter:
prob: 0.5
brightness: 0.1
contrast: 0.1
saturation: 0.1
hue: 0.1
- random_cutout:
n_holes: 4
length: 80
ratio: 0.2
prob: 0.5
random_crop_and_resize:
size:
- 480
- 480
to_tensor: no_params
normalize:
mean: [0.485, 0.456, 0.406]
std: [0.229, 0.224, 0.225]
aug_epoch: 80
val:
bs: 1
augment:
resize:
size:
- 480
- 480
to_tensor: no_params
normalize:
mean: [0.485, 0.456, 0.406]
std: [0.229, 0.224, 0.225]
hyp:
epochs: 100
lr0: 0.001 # sgd=1e-2, adam=1e-3
lrf_ratio: null # decay to lrf_ratio * lr0, if None, 0.1
momentum: 0.937
weight_decay: 0.0005
warmup_momentum: 0.8
warm_ep: 3
loss:
ce: True
bce:
- False
- 0.5
- True
label_smooth: 0.1
strategy:
prog_learn: False
mixup:
- 0.01 # prob
- [0,70] # [start, end)
focal: # Only support BCE
- False # turn-on-or-off
- 0.25 # alpha
- 1.5 # gamma
ohem: # only support CE
- False
- 8 # min_kept
- 0.7 # thresh_prob
- 255 # ignore_index
optimizer:
- sgd # sgd adam or sam
- False
scheduler: cosine_with_warm # linear or cosine
"""
in line 18 and line 19, I set
common_aug: 1 2 3
class_aug: null
It does not work, what I can do to take it to come into effect?
Thanks for your attention!
你好,最新的代码在Oxford上能训练,但是验证和可视化不能实现了?有之前的代码吗?
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.