Negative Predictions in tl_heatmap and br_heatmap for detecting Few Cells

princeton-vl / CornerNet

BSD 3-Clause "New" or "Revised" License

2.36k stars 475 forks source link

Negative Predictions in tl_heatmap and br_heatmap for detecting Few Cells #185

Open Younger330 opened 6 months ago

Younger330 commented 6 months ago

Hello,

I am using CornerNet to train on my dataset for cell detection. My dataset contains images with fewer than ten cells, and I've confirmed that two corner points cover the cells. After training, the predictions for both tl_heatmap and br_heatmap are negative and incorrect. Could you suggest any possible solutions to this issue?

Thank you.