Git Product home page Git Product logo

Comments (2)

bowenc0221 avatar bowenc0221 commented on July 19, 2024

No threshold or NMS is used, we simply keep the top-K instances with the highest scores:

Mask2Former/mask2former/maskformer_model.py

Lines 344 to 380 in 46af009

def instance_inference(self, mask_cls, mask_pred):
# mask_pred is already processed to have the same shape as original input
image_size = mask_pred.shape[-2:]
# [Q, K]
scores = F.softmax(mask_cls, dim=-1)[:, :-1]
labels = torch.arange(self.sem_seg_head.num_classes, device=self.device).unsqueeze(0).repeat(self.num_queries, 1).flatten(0, 1)
# scores_per_image, topk_indices = scores.flatten(0, 1).topk(self.num_queries, sorted=False)
scores_per_image, topk_indices = scores.flatten(0, 1).topk(self.test_topk_per_image, sorted=False)
labels_per_image = labels[topk_indices]
topk_indices = topk_indices // self.sem_seg_head.num_classes
# mask_pred = mask_pred.unsqueeze(1).repeat(1, self.sem_seg_head.num_classes, 1).flatten(0, 1)
mask_pred = mask_pred[topk_indices]
# if this is panoptic segmentation, we only keep the "thing" classes
if self.panoptic_on:
keep = torch.zeros_like(scores_per_image).bool()
for i, lab in enumerate(labels_per_image):
keep[i] = lab in self.metadata.thing_dataset_id_to_contiguous_id.values()
scores_per_image = scores_per_image[keep]
labels_per_image = labels_per_image[keep]
mask_pred = mask_pred[keep]
result = Instances(image_size)
# mask (before sigmoid)
result.pred_masks = (mask_pred > 0).float()
result.pred_boxes = Boxes(torch.zeros(mask_pred.size(0), 4))
# Uncomment the following to get boxes from masks (this is slow)
# result.pred_boxes = BitMasks(mask_pred > 0).get_bounding_boxes()
# calculate average mask prob
mask_scores_per_image = (mask_pred.sigmoid().flatten(1) * result.pred_masks.flatten(1)).sum(1) / (result.pred_masks.flatten(1).sum(1) + 1e-6)
result.scores = scores_per_image * mask_scores_per_image
result.pred_classes = labels_per_image
return result

from mask2former.

tsrobcvai avatar tsrobcvai commented on July 19, 2024

test_topk_per_image
Thank you, what is the default setting for this value <test_topk_per_image>? I am sorry, I didn't find it.

from mask2former.

Related Issues (20)

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.