中文

模型仓库modelzoo

Classification

network	float	qat	quantization	dataset	input shape
mobilenetv1_imagenet	74.12	73.92	73.61	ImageNet	1x3x224x224
mobilenetv2_imagenet	72.65	72.51	72.11	ImageNet	1x3x224x224
resnet18_imagenet	72.04	72.03	72.03	ImageNet	1x3x224x224
resnet50_imagenet	77.37	76.99	76.94	ImageNet	1x3x224x224
vargnetv2_imagenet	73.94	73.56	73.64	ImageNet	1x3x224x224
efficientnet_imagenet	74.31	74.23	74.18	ImageNet	1x3x224x224
horizon_swin_transformer_imagenet	80.24	80.15	80.05	ImageNet	1x3x224x224
mixvargenet_imagenet	71.33	71.23	71.04	ImageNet	1x3x224x224
efficientnasnetm_imagenet	80.24	79.99	79.94	ImageNet	1x3x280x280
efficientnasnets_imagenet	76.63	76.23	76.03	ImageNet	1x3x300x300
vit_small_imagenet	79.50	79.40	-	ImageNet	1x3x224x224
henet_tinye_imagenet	77.68	77.22	-	ImageNet	1x3x224x224
henet_tinym_imagenet	78.38	77.95	-	ImageNet	1x3x224x224

Detection

FCOS

network	backbone	float	qat	quantization	dataset	input shape
fcos_efficientnetb0_mscoco	efficientnetb0	36.26	35.79	35.59	MS COCO	1x3x512x512
fcos_efficientnetb1_mscoco	efficientnetb1	41.37	41.21	40.71	MS COCO	1x3x640x640
fcos_efficientnetb2_mscoco	efficientnetb2	45.35	45.10	45.00	MS COCO	1x3x768x768
fcos_efficientnetb3_mscoco	efficientnetb3	48.03	47.65	47.58	MS COCO	1x3x896x896

DETR

network	backbone	float	qat	quantization	dataset	input shape
detr_resnet50_mscoco	resnet50	35.70	31.42	31.31	MS COCO	1x3x800x1333
detr_efficientnetb3_mscoco	efficientnetb3	37.21	35.95	35.99	MS COCO	1x3x800x1333

Deform DETR

network	backbone	float	qat	quantization	dataset	input shape
deform_detr_resnet50_mscoco	resnet50	44.34	44.65	-	MS COCO	1x3x800x1333

FCOS3D

network	backbone	float	qat	quantization	dataset	input shape
fcos3d_efficientnetb0_nuscenes	efficientnetb0	30.60	30.27	30.31	nuscenes	1x3x512x896

Segmentation

UNet

network	backbone	float	qat	quantization	dataset	input shape
unet_mobilenetv1_cityscapes	MobileNetV1	68.02	67.56	67.53	Cityscapes	1x3x1024x2048

Deeplab

network	backbone	float	qat	quantization	dataset	input shape
deeplabv3plus_efficientnetm0_cityscapes	EfficientNet-M0	76.30	76.22	76.12	Cityscapes	1x3x1024x2048
deeplabv3plus_efficientnetm1_cityscapes	EfficientNet-M1	77.94	77.64	77.65	Cityscapes	1x3x1024x2048
deeplabv3plus_efficientnetm2_cityscapes	EfficientNet-M2	78.82	78.65	78.63	Cityscapes	1x3x1024x2048

FastScnn

network	backbone	float	qat	quantization	dataset	input shape
fastscnn_efficientnetb0tiny_cityscapes	EfficientNet-B0lite	69.97	69.90	69.88	Cityscapes	1x3x1024x2048

OpticalFlow

PwcNet

network	backbone	float	qat	quantization	dataset	input shape
pwcnet_pwcnetneck_flyingchairs	PwcNet	1.4117	1.4112	1.4075	FlyingChairs	1x6x384x512

Lidar

PointPillars

network	backbone	float	qat	quantization	dataset	input shape
pointpillars_kitti_car	SequentialBottleNeck	77.31	76.86	76.76	KITTI3D	150000x4

CenterPoint

network	backbone	float	qat	quantization	dataset	input shape
centerpoint_pointpillar_nuscenes	SequentialBottleNeck	58.32	58.11	58.14	nuscenes	1x5x20x40000, 40000x4

LidarMultiTask

network	backbone	float	qat	quantization	dataset	input shape
centerpoint_mixvargnet_multitask_nuscenes	MixVarGENet	58.09	57.72	57.62	nuscenes	1x5x20x40000, 40000x4

注解

PointPillars 的指标是 Box3d Moderate 这项。

Lane Detection

GaNet

network	backbone	float	qat	quantization	dataset	input shape
ganet_mixvargenet_culane	MixVarGENet	79.49	78.72	78.72	CuLane	1x3x320x800

Multiple Object Track

Motr

network	backbone	float	qat	quantization	dataset	input shape
motr_efficientnetb3_mot17	efficientnetb3	58.02	57.62	57.76	Mot17	1x3x800x1422, 1x256x2x128, 1x1x1x256, 1x4x2x128

Binocular depth estimation

StereoNet

network	backbone	float	qat	quantization	dataset	input shape
stereonet_stereonetneck_sceneflow	StereoNeck	1.1270	1.1677	1.1685	SceneFlow	1x6x540x960
stereonetplus_mixvargenet_sceneflow	MixVarGENet	1.1270	1.1329	1.1351	SceneFlow	2x3x544x960

Bev

Bev

network	backbone	float	qat	quantization	dataset	input shape
bev_ipm_efficientnetb0_multitask_nuscenes	efficientnetb0	30.59	30.80	30.41	nuscenes det	6x3x512x960, 6x128x128x2
bev_ipm_efficientnetb0_multitask_nuscenes	efficientnetb0	51.47	51.41	50.98	nuscenes seg	6x3x512x960, 6x128x128x2
bev_lss_efficientnetb0_multitask_nuscenes	efficientnetb0	30.09	30.05	30.01	nuscenes det	6x3x256x704, 10x128x128x2, 10x128x128x2
bev_lss_efficientnetb0_multitask_nuscenes	efficientnetb0	51.78	51.47	51.46	nuscenes seg	6x3x256x704, 10x128x128x2, 10x128x128x2
bev_gkt_mixvargenet_multitask_nuscenes	MixVarGENet	28.11	28.12	27.90	nuscenes det	6x3x512x960, 6x64x64x2, 6x64x64x2, 6x64x64x2, 6x64x64x2, 6x64x64x2, 6x64x64x2, 6x64x64x2, 6x64x64x2, 6x64x64x2
bev_gkt_mixvargenet_multitask_nuscenes	MixVarGENet	48.53	48.02	48.37	nuscenes seg	6x3x512x960, 6x64x64x2, 6x64x64x2, 6x64x64x2, 6x64x64x2, 6x64x64x2, 6x64x64x2, 6x64x64x2, 6x64x64x2, 6x64x64x2
bev_ipm_4d_efficientnetb0_multitask_nuscenes	efficientnetb0	37.21	37.19	37.17	nuscenes det	6x3x512x960, 6x128x128x2, 1x64x128x128, 1x128x128x2
bev_ipm_4d_efficientnetb0_multitask_nuscenes	efficientnetb0	52.90	53.80	53.77	nuscenes seg	6x3x512x960, 6x128x128x2, 1x64x128x128, 1x128x128x2
detr3d_efficientnetb3_nuscenes	efficientnetb3	34.04	33.87	33.39	nuscenes det	6x3x512x1408
petr_efficientnetb3_nuscenes	efficientnetb3	37.60	37.32	37.31	nuscenes det	6x3x512x1408
bevformer_tiny_resnet50_detection_nuscenes	resnet50	37.00	36.66	-	nuscenes det	6x3x480x800, 1x2500x256, 6x1x2500x4x2, 1x50x50x2
bev_cft_efficientnetb3_nuscenes	efficientnetb3	32.93	32.68	-	nuscenes det	6x3x512x1408
bev_sparse_resnet50_nuscenes	efficientnetb3	56.28	55.23	-	nuscenes det	6x3x256x704, 6x4x4, 1x600x11, 1x600x256, 1x600, 1

Keypoint Detection

HeatmapKeypointModel

network	backbone	float	qat	quantization	dataset	input shape
keypoint_efficientnetb0_carfusion	efficientnetb0	94.33	94.30	94.31	carfusion	1x3x128x128

Trajectory Prediction

DenseTNT

network	backbone	float	qat	quantization	dataset	input shape
densetnt_vectornet_argoverse1	vectornet	1.2974	1.2989	1.3038	argoverse 1	30x9x19x32, 30x11x9x64, 30x1x1x96, 30x2x1x2048, 30x1x1x2048

QCNet

network	backbone	float	qat	quantization	dataset	input shape
qcnet_oe_argoverse2	-	83.84	83.26	-	argoverse 2	输入见下方list

注解

qcnet_oe_argoverse2 的指标是 HitRate 这项。

qcnet_oe_argoverse2 模型输入shape为:

1x30x50, 1x50x30x30, 1x30x1, 1x1x30x1, 1x1x30x1, 1x1x30x1, 1x1x30x1, 1x1x30x80, 1x1x30x80, 1x1x30x80, 1x1x30x10, 1x1x30x10, 1x1x30x10, 1x1x30x10, 1x1x30x30, 1x1x30x30, 1x1x30x30, 1x30x29x128, 1x30x10x128, 1x30x10x128, 1x80, 1x80, 1x1x80x80, 1x1x80x80, 1x1x80x80, 1x1x80x50, 1x1x80x50, 1x1x80x50, 1x80x50, 1x80x50, 1x80x50, 1x80x50, 1x30x30, 1x30x1, 1x1x30x30, 1x1x30x30, 1x1x30x30, 1x1x30x30, 1x1x30x80, 1x1x30x80, 1x1x30x80, 1x1x30x30, 1x1x30x30, 1x1x30x30, 1x80x80.

Occupancy Prediction

FlashOcc

network	backbone	float	qat	quantization	dataset	input shape
flashocc_henet_lss_occ3d_nuscenes	henet_tinym_imagenet	0.3646	0.3622	-	occ3d_nuscenes	6x3x512x960, 10x128x128x2, 10x128x128x2

Online Map Construction

MapTR

network	backbone	float	qat	quantization	dataset	input shape
maptrv2_resnet50_bevformer_nuscenes	resnet50	0.5859	0.5843	0.5763	nuscenes	6x3x480x800, 6x1x2500x4x2

页面目录

hb_compile工具

QAT

模型导出

Horizon算子

视觉算法

示例

VP API手册

数据结构

功能接口

模型推理API手册

数据类型和数据结构

API接口

模型推理工具介绍

高性能算法

HPL API手册

数据结构

功能接口

数据结构

功能接口

使用框架

使用指南

示例说明

API手册

#模型仓库modelzoo

#Classification

#Detection

#Segmentation

#OpticalFlow

#Lidar

#Lane Detection

#Multiple Object Track

#Binocular depth estimation

#Bev

#Keypoint Detection

#Trajectory Prediction

#Occupancy Prediction

#Online Map Construction

模型仓库modelzoo

Classification

Detection

Segmentation

OpticalFlow

Lidar

Lane Detection

Multiple Object Track

Binocular depth estimation

Bev

Keypoint Detection

Trajectory Prediction

Occupancy Prediction

Online Map Construction