Skip to content

Latest commit

 

History

History
141 lines (108 loc) · 5.42 KB

validation_results_ncnn.md

File metadata and controls

141 lines (108 loc) · 5.42 KB
Raw

Validation results for the models inferring using ncnn

Image classification

Test image #1

Data source: ImageNet

Image resolution: 709 x 510 

Model Python API
shufflenetv2 0.5396927 piggy bank, penny bank
0.0453512 saltshaker, salt shaker
0.0443007 whistle
0.0347720 ocarina, sweet potato
0.0286027 lemon
squeezenet 0.9628906 Granny Smith
0.0068016 lemon
0.0064964 fig
0.0046844 tennis ball
0.0038204 piggy bank, penny bank

Test image #2

Data source: ImageNet

Image resolution: 500 x 500 

Model Python API
shufflenetv2 0.9906778 junco, snowbird
0.0034630 brambling, Fringilla montifringilla
0.0023069 house finch, linnet, Carpodacus mexicanus
0.0017143 chickadee
0.0006609 goldfinch, Carduelis carduelis
squeezenet 0.9804688 junco, snowbird
0.0173798 chickadee
0.0005875 jay
0.0003612 indigo bunting, indigo finch, indigo bird, Passerina cyanea
0.0003293 brambling, Fringilla montifringilla

Test image #3

Data source: ImageNet

Image resolution: 333 x 500 

Model Python API
shufflenetv2 0.2229400 breakwater, groin, groyne, mole, bulwark, seawall, jetty
0.2029892 liner, ocean liner
0.0577048 fireboat
0.0493575 dock, dockage, docking facility
0.0428826 container ship, containership, container vessel
squeezenet 0.8725586 lifeboat
0.0500183 container ship, containership, container vessel
0.0284729 drilling platform, offshore rig
0.0120697 pirate, pirate ship
0.0110016 dock, dockage, docking facility

Object detection

Test image #1

Data source: ImageNet

Image resolution: 500 x 500 

Bounding box (upper left and bottom right corners):
(117, 86), (365, 465)
Model Python API
faster_rcnn Bounding box:
(58, 141), (359, 484)
mobilenet_ssd Bounding box:
(94, 93), (359, 481)
mobilenetv2_ssdlite Bounding box:
(76, 100), (347, 460)
mobilenetv3_ssdlite Bounding box:
(61, 86), (365, 498)
mobilenet_yolov2 Bounding box:
(72, 101), (341, 466)
mobilenetv2_yolov3 Bounding box:
(84, 92), (354, 473)
rfcn Bounding box:
(93, 99), (334, 445)
squeezenet_ssd Bounding box:
(98, 103), (350, 449)
yolov4_tiny Bounding box:
(74, 85), (243, 425)
yolov5s Bounding box:
(68, 96), (355, 490)
yolov8s Bounding box:
(59, 100), (352, 447)

Test image #2

Data source: ImageNet

Image resolution: 333 x 500 

Bounding box (upper left and bottom right corners):
(82, 262), (269, 376)
Model Python API
faster_rcnn Bounding box:
(58, 180), (282, 418)
mobilenet_ssd Bounding box:
(79, 140), (270, 375)
mobilenetv2_ssdlite Bounding box:
(82, 265), (267, 376)
mobilenetv3_ssdlite Bounding box:
(59, 112), (295, 414)
mobilenet_yolov2 Bounding box:
(54, 139), (277, 375)
mobilenetv2_yolov3 Bounding box:
(75, 127), (276, 390)
rfcn Bounding box:
(88, 138), (259, 381)
squeezenet_ssd Bounding box:
(78, 149), (260, 357)
yolov4_tiny Bounding box:
(96, 265), (244, 371)
yolov5s Bounding box:
(81, 249), (267, 377)
yolov8s Bounding box:
(82, 242), (269, 378)

Test image #3

Data source: Pascal VOC

Image resolution: 500 x 375 

Bounding box (upper left and bottom right corners):
(62, 127), (443, 251)
Model Python API
faster_rcnn Bounding box:
(6, 94), (477, 257)
mobilenet_ssd Bounding box:
(54, 128), (447, 244)
mobilenetv2_ssdlite Bounding box:
(61, 128), (435, 238)
mobilenetv3_ssdlite Bounding box:
(37, 105), (295, 414)
mobilenet_yolov2 Bounding box:
(59, 112), (433, 239)
mobilenetv2_yolov3 Bounding box:
(62, 124), (427, 241)
rfcn Bounding box:
(46, 102), (436, 252)
squeezenet_ssd Bounding box:
(47, 118), (458, 248)
yolov4_tiny Bounding box:
(55, 124), (427, 241)
yolov5s Bounding box:
(41, 118), (441, 245)
yolov8s Bounding box:
(58, 122), (434, 243)