Inference - MindOCR Models
MindOCR Support Models List
Note: All results is test on 310P3 with MindSpore2.2.14.
Text Detection
Model |
Backbone |
Language |
Datset |
F-score(%) |
FPS |
Data Shape (NCHW) |
Lite convert config txt |
Configuration File |
Download Link |
DBNet |
MobileNetV3 |
en |
IC15 |
76.96 |
24.90 |
(1,3,736,1280) |
config.txt |
yaml |
mindir |
|
ResNet-18 |
en |
IC15 |
81.73 |
24.16 |
(1,3,736,1280) |
config.txt |
yaml |
mindir |
|
ResNet-50 |
en |
IC15 |
85.00 |
25.98 |
(1,3,736,1280) |
config.txt |
yaml |
mindir |
|
ResNet-50 |
ch + en |
/ |
/ |
/ |
(1,3,736,1280) |
config.txt |
yaml |
mindir |
DBNet++ |
ResNet-50 |
en |
IC15 |
86.79 |
6.75 |
(1,3,1152,2048) |
config.txt |
yaml |
mindir |
|
ResNet-50 |
ch + en |
/ |
/ |
/ |
(1,3,1152,2048) |
config.txt |
yaml |
mindir |
EAST |
ResNet-50 |
en |
IC15 |
84.18 |
23.52 |
(1,3,720,1280) |
config.txt |
yaml |
mindir |
|
MobileNetV3 |
en |
IC15 |
75.08 |
24.37 |
(1,3,720,1280) |
config.txt |
yaml |
mindir |
PSENet |
ResNet-152 |
en |
IC15 |
81.69 |
2.94 |
(1,3,1472,2624) |
config.txt |
yaml |
mindir |
|
ResNet-50 |
en |
IC15 |
81.36 |
9.94 |
(1,3,736,1312) |
config.txt |
yaml |
mindir |
|
MobileNetV3 |
en |
IC15 |
70.67 |
10.33 |
(1,3,736,1312) |
config.txt |
yaml |
mindir |
FCENet |
ResNet50 |
en |
IC15 |
77.99 |
16.97 |
(1,3,736,1280) |
config.txt |
yaml |
mindir |
Text Recognition
Model |
Backbone |
Character Dict |
Dataset |
Acc(%) |
FPS |
Data Shape (NCHW) |
Lite convert config txt |
Configuration File |
Download Link |
CRNN |
VGG7 |
Default |
IC15 |
66.01 |
394.30 |
(1,3,32,100) |
config.txt |
yaml |
mindir |
|
ResNet34_vd |
Default |
IC15 |
69.67 |
339.45 |
(1,3,32,100) |
config.txt |
yaml |
mindir |
|
ResNet34_vd |
ch_dict.txt |
/ |
/ |
/ |
(1,3,32,320) |
config.txt |
yaml |
mindir |
SVTR |
Tiny |
Default |
IC15 |
80.02 |
314.08 |
(1,3,64,256) |
config.txt |
yaml |
mindir |
Rare |
ResNet34_vd |
Default |
IC15 |
69.48 |
239.66 |
(1,3,32,100) |
config.txt |
yaml |
mindir |
|
ResNet34_vd |
ch_dict.txt |
/ |
/ |
/ |
(1,3,32,320) |
config.txt |
yaml |
mindir |
RobustScanner |
ResNet-31 |
en_dict90.txt |
IC15 |
78.62 |
63.81 |
(1,3,48,160) |
config.txt |
yaml |
mindir |
VisionLAN |
ResNet-45 |
Default |
IC15 |
80.07 |
301.49 |
(1,3,64,256) |
config.txt |
yaml(LA) |
mindir(LA) |
Text Direction Classification
Model |
Backbone |
Dataset |
Acc(%) |
FPS |
Data Shape (NCHW) |
Lite convert config txt |
Configuration File |
Download Link |
MobileNetV3 |
MobileNetV3 |
/ |
/ |
/ |
(1,3,48,192) |
config.txt |
yaml |
mindir |