Note: We follow the guideline provided by MeshTransformer/docs/DEMO.md

Demo

We provide guidelines to run end-to-end inference on test images.

Our code iterates all images in the given folder, and produces estimation results.

• 3D Human Mesh Reconstruction
• 3D Hand Mesh Reconstruction

---

3D Human Mesh Reconstruction

This demo runs 3D human mesh reconstruction from a single RGB image.

Our codes require the input images that are already cropped with the person centered in the image. The input images should have the size of 224x224. To run the demo, please place your test images under ./demo/human-body, and then run the following script. After running, it will generate the results in the folder ./demo/human-body.

python ./src/tools/end2end_inference_bodymesh.py \
       --resume_checkpoint ./models/fastmetro_checkpoint/FastMETRO-L-H64_3dpw_state_dict.bin \
       --image_file_or_path ./demo/human-body 

• To use SMPL parameter regressor, you need to set --use_smpl_param_regressor as True

Demo Result

• obtained by using the released model checkpoint (FastMETRO-L-H64_3dpw_state_dict.bin)

• obtained by using the released model checkpoint (FastMETRO-L-H64_smpl_3dpw_state_dict.bin)

• (Left) Input Image
• (Middle) FastMETRO
• (Right) FastMETRO with an optional SMPL parameter regressor

3D Hand Mesh Reconstruction

This demo runs 3D hand mesh reconstruction from a single RGB image.

Our codes require the input images that are already cropped with the right-hand centered in the image. The input images should have the size of 224x224. To run the demo, please place your test images under ./demo/hand, and then run the following script. After running, it will generate the results in the folder ./demo/hand.

For the left-hand centered images, please horizontally flip the images before running this demo

python ./src/tools/end2end_inference_handmesh.py \
       --resume_checkpoint ./models/fastmetro_checkpoint/FastMETRO-L-H64_freihand_state_dict.bin \
       --image_file_or_path ./demo/hand

Demo Result

• obtained by using the released model checkpoint (FastMETRO-L-H64_freihand_state_dict.bin)