Basic idea: The project needs a TensorRT model for detecting human attributes. I searched for one and turned it around. It is relatively simple. I deploy it on the tensorRT platform and then integrate it into monocular 3D target detection.
Test pictures, experimental model link: https://pan.baidu.com/s/1h06cSe-_2zOngViC6lMr4w?pwd=mwh1 Extraction code: mwh1
1. Download source code and test
ubuntu@ubuntu:~$ pip3 install paddleclas
ubuntu@ubuntu:~$ python3 -m pip install paddlepaddle-gpu -i https://mirror.baidu.com/pypi/simple
ubuntu@ubuntu:~$ paddleclas --model_name=person_attribute --infer_imgs=/home/ubuntu/PaddleClas/deploy/images/Pedestrain_Attr.jpg
[2023/07/20 12:34:28] ppcls INFO: attributes: ['Male', 'Age18-60', 'Front', 'Glasses: True', 'Hat: False', 'HoldObjectsInFront: False', 'No bag', 'Upper: LongSleeve UpperPlaid', 'Lower: Trousers', 'No boots'], output: [0, 1, 0, 1, 0, 0, 1, 0, 0, 0, 0, 1,