ILSVRC2012

All the classification networks are pre-trained on ILSVRC2012 and the accuracy is measured on the 50k-image validation set.
ModelFull Precision Acc (top1)Quantized Acc (top1)Input Resolution (HxWxC)Params (M)Flops (G)Pretrained ModelCompiled ModelSource
efficientnet_l80.4780.07300x300x39.019.68pretrainedcompiledlink
efficientnet_m78.9878.47240x240x35.583.67pretrainedcompiledlink
efficientnet_s77.6177.11224x224x34.132.36pretrainedcompiledlink
efficientnet_lite074.9374.15224x224x33.350.39pretrainedcompiledlink
efficientnet_lite176.6676.16240x240x34.110.6pretrainedcompiledlink
efficientnet_lite277.4476.3260x260x34.780.86pretrainedcompiledlink
efficientnet_lite379.277.99280x280x36.881.39pretrainedcompiledlink
efficientnet_lite480.6479.55300x300x311.672.56pretrainedcompiledlink
inception_v169.7669.41224x224x35.591.5pretrainedcompiledlink
mobilenet_v171.0269.96224x224x33.20.57pretrainedcompiledlink
mobilenet_v2_1.071.8469.95224x224x32.210.31pretrainedcompiledlink
mobilenet_v2_1.474.1173.2224x224x34.290.59pretrainedcompiledlink
mobilenet_v372.2771.87224x224x32.791pretrainedcompiledlink
mobilenet_v3_large_minimalistic72.2969.44224x224x31.40.21pretrainedcompiledlink
regnetx_1.6gf77.0776.65224x224x38.261.61pretrainedcompiledlink
regnetx_800mf75.0774.57224x224x36.570.8pretrainedcompiledlink
regnety_200mf70.3269.48224x224x33.150.2pretrainedcompiledlink
resnet_v1_1868.8468.51224x224x311.171.82pretrainedcompiledlink
resnet_v1_3472.6871.59224x224x321.283.67pretrainedcompiledlink
resnet_v1_50^75.2174.85224x224x323.483.49pretrainedcompiledlink
resnet_v2_1869.5868.01224x224x311.171.82pretrainedcompiledlink
resnet_v2_3473.172.77224x224x321.283.67pretrainedcompiledlink
resnext26_32x4d76.0874.84224x224x313.332.51pretrainedcompiledlink
resnext50_32x4d79.4378.55224x224x322.964.29pretrainedcompiledlink
shufflenet_g8_w166.2965.56224x224x30.920.18pretrainedcompiledlink
squeezenet_v1.159.8859.18224x224x31.240.39pretrainedcompiledlink

^ networks available in TAPPAS