Add export module #1
Draft
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| @@ -120,14 +120,18 @@ def forward(self, images, annotations = None): | |||
| #jloc_pred_nms = non_maximum_suppression(jloc_pred[0]) | |||
| #topK = torch.clamp((jloc_pred_nms > 0.008).count_nonzero(), max=300) | |||
|
|
|||
| juncs_pred, _ = get_junctions(non_maximum_suppression(jloc_pred[0]),joff_pred[0], topk=300, th=0.008) | |||
| nms_jloc_pred = non_maximum_suppression(jloc_pred)[0] | |||
There was a problem hiding this comment.
This is the part where I pass the 4 dimensional tensor, and then take out the batch dim after max_pool2d
Comment on lines
+130
to
+128
| # idx_junc_to_end_min = torch.min(idx_junc_to_end1,idx_junc_to_end2) | ||
| # idx_junc_to_end_max = torch.max(idx_junc_to_end1,idx_junc_to_end2) |
There was a problem hiding this comment.
torch.min and torch.max signatures are (tensor, dim, keep_dims, ..)
They are used here are torch.min/max(tensor1, tensor2) to do element-wise minimum/maximum operations.
This does not work for onnx, as it sees min and max, and expects the second argument to be an integer, and not a tensor.
The fix is to stack the two tensors together, then take the min/max along the stacked dimension.
| wireframe = WireframeGraph(juncs_final,juncs_score,edge_indices,score_final,output.size(3),output.size(2)) | ||
| wireframe.rescale(annotations[0]['width'],annotations[0]['height']) | ||
| else: | ||
| return juncs_final,juncs_score,edge_indices,score_final,output.size(3),output.size(2) |
There was a problem hiding this comment.
output.size(3) and output.size(2) are not tensors anymore, they probably won't have meaningful data once exported
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Adds a module that exports the model to the ONNX framework and executes a test run.
Some minor modifications are made to the model to make it exportable but should not affect the outputs.