We provide the config files for training on different dataset combination:
- Mix 1: H36M, MI, COCO
- Mix 2: H36M, MI, EFT-COCO
- Mix 3: H36M, MI, EFT-COCO, MPII
- Mix 4: H36M, MuCo, EFT-COCO
- Mix 5: H36M, MI, COCO, LSP, LSPET, MPII
- Mix 6: EFT-[COCO, MPII, LSPET], SPIN-MI, H36M
- Mix 7: EFT-[COCO, MPII, LSPET], MuCo, H36M, PROX
- Mix 8: EFT-[COCO, PT, LSPET], MI, H36M
- Mix 9: EFT-[COCO, PT, LSPET, OCH], MI, H36M
- Mix 10: PROX, MuCo, EFT-[COCO, PT, LSPET, OCH], UP-3D, MTP, Crowdpose
- Mix 11: EFT-[COCO, MPII, LSPET], MuCo, H36M
We evaluate trained models on 3DPW. Values are MPJPE/PA-MPJPE.
| Mixes | Datasets Config | 3DPW | |:------:|:------:|:-------:|:-------:| | Mix 1 | H36M, MI, COCO | resnet50_hmr_mix1.py | 66.14 | | Mix 2 | H36M, MI, EFT-COCO |resnet50_hmr_mix2.py | 55.98 | | Mix 3 | H36M, MI, EFT-COCO, MPII |resnet50_hmr_mix3.py | 56.12 | | Mix 4 | H36M, MuCo, EFT-COCO |resnet50_hmr_mix4.py | 53.90 | | Mix 5 | H36M, MI, COCO, LSP, LSPET, MPII |resnet50_hmr_mix5.py | 64.55 | | Mix 6 | EFT-[COCO, MPII, LSPET], SPIN-MI, H36M |resnet50_hmr_mix6.py | 55.47 | | Mix 7 | EFT-[COCO, MPII, LSPET], MuCo, H36M, PROX |resnet50_hmr_mix7.py | 53.44 | | Mix 8 | EFT-[COCO, PT, LSPET], MI, H36M |resnet50_hmr_mix8.py | 55.97 | | Mix 9 | EFT-[COCO, PT, LSPET, OCH], MI, H36M |resnet50_hmr_mix9.py | 55.59 | | Mix 10 | PROX, MuCo, EFT-[COCO, PT, LSPET, OCH], UP-3D, MTP, Crowdpose |resnet50_hmr_mix10.py | 57.84 | | Mix 11 | EFT-[COCO, MPII, LSPET], MuCo, H36M |resnet50_hmr_mix11.py | 52.54 |