search

lucasjinreal / yolov7_d2

🔥🔥🔥🔥 (Earlier YOLOv7 not official one) YOLO with Transformers and Instance Segmentation, with TensorRT acceleration! 🔥🔥🔥
GNU General Public License v3.0
3.13k stars 483 forks source link

not getting proper bounding box for head detection #216

Open sathyanarayanmc opened 1 year ago

sathyanarayanmc commented 1 year ago

I tried some custom head dataset ,but model is not giving proper bounding box like its plotting half head,so what will be the problem..

brmarkus commented 1 year ago

Can you provide more details, please?

What do you mean with "tried some custom head dataset"? You retrained the YoloV7-D2 model with your dataset?

What does it look like, when you say "plotting half head"?

From the repo: https://camo.githubusercontent.com/e6474c64819243473b8c6964a4d9d4b2a1f56d7c19592c55159278b0aef3088c/68747470733a2f2f7a332e617831782e636f6d2f323032312f30372f32322f5744723556302e706e67 image Do you mean something like this, "half head" (because the head is rotated)?

lucasjinreal commented 1 year ago

You can try yolox head if you want which might won't have anchor issues on your dataset.