Loss Functions
Dynamic SmoothL1 Loss
Introduced by Zhang et al. in Dynamic R-CNN: Towards High Quality Object Detection via Dynamic TrainingDynamic SmoothL1 Loss (DSL) is a loss function in object detection where we change the shape of loss function to gradually focus on high quality samples:
$$\text{DSL}\left(x, \beta_{now}\right) = 0.5|{x}|^{2}/\beta_{now}, \text{ if } |x| < \beta_{now}\text{,} $$ $$\text{DSL}\left(x, \beta_{now}\right) = |{x}| - 0.5\beta_{now}\text{, otherwise} $$
DSL will change the value of $\beta_{now}$ according to the statistics of regression errors which can reflect the localization accuracy. It was introduced as part of the Dynamic R-CNN model.
Source: Dynamic R-CNN: Towards High Quality Object Detection via Dynamic Training
Papers
| Paper | Code | Results | Date | Stars |
|---|
Tasks
| Task | Papers | Share |
|---|---|---|
| Object Detection | 3 | 33.33% |
| Fire Detection | 1 | 11.11% |
| Management | 1 | 11.11% |
| Ensemble Learning | 1 | 11.11% |
| Fracture detection | 1 | 11.11% |
| Medical Object Detection | 1 | 11.11% |
| regression | 1 | 11.11% |
Usage Over Time
Components
| Component | Type |
|
|---|---|---|
| 🤖 No Components Found | You can add them if they exist; e.g. Mask R-CNN uses RoIAlign |
