top_down_img_demo_with_mmdet.py

[hsp2454]

In the det_checkpoint file top_down_img_demo_with_mmdet.py, if I want to train my own data, then in addition to training the attitude estimation model, I also need to train a target detection model in advance to get the weight file

[Ben-Louis]

Hi, if your target objects are included in COCO category, you can directly use detectors trained on COCO such as the one in the following command

python demo/top_down_img_demo_with_mmdet.py \
    demo/mmdetection_cfg/faster_rcnn_r50_fpn_coco.py \
    https://download.openmmlab.com/mmdetection/v2.0/faster_rcnn/faster_rcnn_r50_fpn_1x_coco/faster_rcnn_r50_fpn_1x_coco_20200130-047c8118.pth \
    ...

Otherwise, you may need to train an object detector with MMDetection on your data.