Views
No views yet
fasterrcnn_resnet50_fpn.| File | Role |
|---|---|
fasterrcnn_resnet50_fpn_backbone_body_dynamic_hw.tflite | ResNet backbone body: transformed image tensor to C2-C5 feature maps |
fasterrcnn_resnet50_fpn_rpn_head_dynamic_hw.tflite | RPN head: one FPN level to objectness logits and box deltas |
fasterrcnn_resnet50_fpn_roi_box_dynamic_n.tflite | ROI box head and predictor: pooled ROI features to class logits and box deltas |
pip install ai-edge-litert torch torchvision pillow numpypython sample_torchvision_fasterrcnn_litert_cpu.py.tflite files in the same directory as the
sample. You can override them explicitly:1python sample_torchvision_fasterrcnn_litert_cpu.py \
2 --backbone-model fasterrcnn_resnet50_fpn_backbone_body_dynamic_hw.tflite \
3 --rpn-head-model fasterrcnn_resnet50_fpn_rpn_head_dynamic_hw.tflite \
4 --roi-model fasterrcnn_resnet50_fpn_roi_box_dynamic_n.tflite \
5 --image https://github.com/pytorch/hub/raw/master/images/dog.jpg1input image: original=(1213, 1546) transformed=(1, 3, 800, 1024)
2host proposal decode/NMS: 1000 proposals
3ROI LiteRT outputs: logits=(1000, 91) box_regression=(1000, 364)
4detections above 0.50: 1
5 01: dog score=0.9669 box=[137.84, 67.8, 1386.9, 1172.82]HardwareAccelerator.CPU.1@article{DBLP:journals/corr/RenHG015,
2 author = {Shaoqing Ren and
3 Kaiming He and
4 Ross B. Girshick and
5 Jian Sun},
6 title = {Faster {R-CNN:} Towards Real-Time Object Detection with Region Proposal
7 Networks},
8 journal = {CoRR},
9 volume = {abs/1506.01497},
10 year = {2015},
11 url = {http://arxiv.org/abs/1506.01497},
12 eprinttype = {arXiv},
13 eprint = {1506.01497},
14 timestamp = {Mon, 13 Aug 2018 16:46:02 +0200},
15 biburl = {https://dblp.org/rec/journals/corr/RenHG015.bib},
16 bibsource = {dblp computer science bibliography, https://dblp.org}
17}