Use torchmetrics to calculate map #99
Conversation
|
Thank you for your contribution! Next, I will proceed with a code review for each block. Please direct push this to the Best regards, |
|
I would just like to mention that torch metrics[detection] is a wrapper around pycocotools SEE THIS
|
| @@ -134,16 +134,16 @@ def load_valid_labels(self, label_path: str, seg_data_one_img: list) -> Union[Te | |||
| def get_data(self, idx): | |||
There was a problem hiding this comment.
Could you add return types please?
| @@ -134,16 +134,16 @@ def load_valid_labels(self, label_path: str, seg_data_one_img: list) -> Union[Te | |||
| def get_data(self, idx): | |||
| img_path, bboxes = self.data[idx] | |||
| img = Image.open(img_path).convert("RGB") | |||
| return img, bboxes, img_path | |||
| return img, bboxes | |||
|
|
|||
| def get_more_data(self, num: int = 1): | |||
There was a problem hiding this comment.
Could you add return types please?
| @@ -227,7 +227,19 @@ def __call__(self, target: Tensor, predict: Tuple[Tensor]) -> Tuple[Tensor, Tens | |||
| 2. Select the targets | |||
| 2. Noramlize the class probilities of targets | |||
There was a problem hiding this comment.
| 2. Noramlize the class probilities of targets | |
| 2. Normalize the class probabilities of targets |
| ap_table.add_row(f"{epoch: 3d}", ap_name, f"{ap_color}{ap_value:.2f}", ar_name, f"{ar_color}{ar_value:.2f}") | ||
|
|
||
| return ap_table, this_ap | ||
| def format_prediction(prediction: torch.Tensor) -> dict: |
There was a problem hiding this comment.
I think for better typing you should better annotate the return type here. Something like below (not exact) to provide at least a little bit better IntelliSense to any contributors:
| def format_prediction(prediction: torch.Tensor) -> dict: | |
| def format_prediction(prediction: torch.Tensor) -> Dict[str, List[int]]: |
| def format_prediction(prediction: torch.Tensor) -> dict: | ||
| """ | ||
| Format the prediction output to a dictionary. | ||
| Args: |
No description provided.