Tencent
rtdetr-l转ncnn
error log | 日志或报错信息 | ログ
model | 模型 | モデル
- rtdetr-l model 下载路径: https://github.com/ultralytics/assets/releases/download/v8.2.0/rtdetr-l.pt
how to reproduce | 复现步骤 | 再現方法
1.rtdetr-l.pt 2.训练后产生：best.pt 3.直接导出ncnn
PyTorch: starting from 'best.pt' with input shape (1, 3, 640, 640) BCHW and output shape(s) (1, 300, 5) (63.1 MB)
TorchScript: starting export with torch 2.4.0+cu121... TorchScript: export success ✅ 12.9s, saved as 'best.torchscript' (122.8 MB)
NCNN: starting export with NCNN 1.0.20240410... NCNN: running '/usr/local/lib/python3.9/site-packages/ultralytics/pnnx best.torchscript ncnnparam=best_ncnn_model/model.ncnn.param ncnnbin=best_ncnn_model/model.ncnn.bin ncnnpy=best_ncnn_model/model_ncnn.py pnnxparam=best_ncnn_model/model.pnnx.param pnnxbin=best_ncnn_model/model.pnnx.bin pnnxpy=best_ncnn_model/model_pnnx.py pnnxonnx=best_ncnn_model/model.pnnx.onnx fp16=0 device=cpu inputshape="[1, 3, 640, 640]"' pnnxparam = best_ncnn_model/model.pnnx.param pnnxbin = best_ncnn_model/model.pnnx.bin pnnxpy = best_ncnn_model/model_pnnx.py pnnxonnx = best_ncnn_model/model.pnnx.onnx ncnnparam = best_ncnn_model/model.ncnn.param ncnnbin = best_ncnn_model/model.ncnn.bin ncnnpy = best_ncnn_model/model_ncnn.py fp16 = 0 optlevel = 2 device = cpu inputshape = [1,3,640,640]f32 inputshape2 = customop = moduleop = ############# pass_level0 inline module = torch.nn.modules.linear.Identity inline module = ultralytics.nn.modules.block.HGBlock inline module = ultralytics.nn.modules.block.HGStem inline module = ultralytics.nn.modules.block.RepC3 inline module = ultralytics.nn.modules.conv.Concat inline module = ultralytics.nn.modules.conv.Conv inline module = ultralytics.nn.modules.conv.DWConv inline module = ultralytics.nn.modules.conv.LightConv inline module = ultralytics.nn.modules.conv.RepConv inline module = ultralytics.nn.modules.head.RTDETRDecoder inline module = ultralytics.nn.modules.transformer.AIFI inline module = ultralytics.nn.modules.transformer.DeformableTransformerDecoder inline module = ultralytics.nn.modules.transformer.DeformableTransformerDecoderLayer inline module = ultralytics.nn.modules.transformer.MLP inline module = ultralytics.nn.modules.transformer.MSDeformAttn inline module = torch.nn.modules.linear.Identity inline module = ultralytics.nn.modules.block.HGBlock inline module = ultralytics.nn.modules.block.HGStem inline module = ultralytics.nn.modules.block.RepC3 inline module = ultralytics.nn.modules.conv.Concat inline module = ultralytics.nn.modules.conv.Conv inline module = ultralytics.nn.modules.conv.DWConv inline module = ultralytics.nn.modules.conv.LightConv inline module = ultralytics.nn.modules.conv.RepConv inline module = ultralytics.nn.modules.head.RTDETRDecoder inline module = ultralytics.nn.modules.transformer.AIFI inline module = ultralytics.nn.modules.transformer.DeformableTransformerDecoder inline module = ultralytics.nn.modules.transformer.DeformableTransformerDecoderLayer inline module = ultralytics.nn.modules.transformer.MLP inline module = ultralytics.nn.modules.transformer.MSDeformAttn
inline module = torch.nn.modules.linear.Identity inline module = ultralytics.nn.modules.block.HGBlock inline module = ultralytics.nn.modules.block.HGStem inline module = ultralytics.nn.modules.block.RepC3 inline module = ultralytics.nn.modules.conv.Concat inline module = ultralytics.nn.modules.conv.Conv inline module = ultralytics.nn.modules.conv.DWConv inline module = ultralytics.nn.modules.conv.LightConv inline module = ultralytics.nn.modules.conv.RepConv inline module = ultralytics.nn.modules.head.RTDETRDecoder inline module = ultralytics.nn.modules.transformer.AIFI inline module = ultralytics.nn.modules.transformer.DeformableTransformerDecoder inline module = ultralytics.nn.modules.transformer.DeformableTransformerDecoderLayer inline module = ultralytics.nn.modules.transformer.MLP inline module = ultralytics.nn.modules.transformer.MSDeformAttn
############# pass_level1 ############# pass_level2 ############# pass_level3 ############# pass_level4 ############# pass_level5 ############# pass_ncnn force batch axis 233 for operand 240 force batch axis 233 for operand 241 binaryop tensor0 with rank 6 is not supported yet! binaryop tensor1 with rank 6 is not supported yet! binaryop tensor0 with rank 6 is not supported yet! binaryop tensor1 with rank 6 is not supported yet! binaryop tensor0 with rank 6 is not supported yet! binaryop tensor1 with rank 6 is not supported yet! binaryop tensor0 with rank 6 is not supported yet! binaryop tensor1 with rank 6 is not supported yet! binaryop tensor0 with rank 6 is not supported yet! binaryop tensor1 with rank 6 is not supported yet! binaryop tensor0 with rank 6 is not supported yet! binaryop tensor1 with rank 6 is not supported yet! binaryop tensor0 with rank 6 is not supported yet! binaryop tensor1 with rank 6 is not supported yet! binaryop tensor0 with rank 6 is not supported yet! binaryop tensor1 with rank 6 is not supported yet! binaryop tensor0 with rank 6 is not supported yet! binaryop tensor1 with rank 6 is not supported yet! binaryop tensor0 with rank 6 is not supported yet! binaryop tensor1 with rank 6 is not supported yet! binaryop tensor0 with rank 6 is not supported yet! binaryop tensor1 with rank 6 is not supported yet! binaryop tensor0 with rank 6 is not supported yet! binaryop tensor1 with rank 6 is not supported yet! munmap_chunk(): invalid pointer NCNN: export failure ❌ 39.0s: Command '['/usr/local/lib/python3.9/site-packages/ultralytics/pnnx', 'best.torchscript', 'ncnnparam=best_ncnn_model/model.ncnn.param', 'ncnnbin=best_ncnn_model/model.ncnn.bin', 'ncnnpy=best_ncnn_model/model_ncnn.py', 'pnnxparam=best_ncnn_model/model.pnnx.param', 'pnnxbin=best_ncnn_model/model.pnnx.bin', 'pnnxpy=best_ncnn_model/model_pnnx.py', 'pnnxonnx=best_ncnn_model/model.pnnx.onnx', 'fp16=0', 'device=cpu', 'inputshape="[1, 3, 640, 640]"']' died with <Signals.SIGABRT: 6>. Traceback (most recent call last): File "/data/train_table/DETR/table/train2/weights/convert.py", line 8, in model.export(format="ncnn") File "/usr/local/lib/python3.9/site-packages/ultralytics/engine/model.py", line 591, in export return Exporter(overrides=args, _callbacks=self.callbacks)(model=self.model) File "/usr/local/lib/python3.9/site-packages/torch/utils/_contextlib.py", line 116, in decorate_context return func(*args, **kwargs) File "/usr/local/lib/python3.9/site-packages/ultralytics/engine/exporter.py", line 325, in call f[11], _ = self.export_ncnn() File "/usr/local/lib/python3.9/site-packages/ultralytics/engine/exporter.py", line 142, in outer_func raise e File "/usr/local/lib/python3.9/site-packages/ultralytics/engine/exporter.py", line 137, in outer_func f, model = inner_func(*args, **kwargs) File "/usr/local/lib/python3.9/site-packages/ultralytics/engine/exporter.py", line 594, in export_ncnn subprocess.run(cmd, check=True) File "/usr/local/lib/python3.9/subprocess.py", line 524, in run raise CalledProcessError(retcode, process.args, subprocess.CalledProcessError: Command '['/usr/local/lib/python3.9/site-packages/ultralytics/pnnx', 'best.torchscript', 'ncnnparam=best_ncnn_model/model.ncnn.param', 'ncnnbin=best_ncnn_model/model.ncnn.bin', 'ncnnpy=best_ncnn_model/model_ncnn.py', 'pnnxparam=best_ncnn_model/model.pnnx.param', 'pnnxbin=best_ncnn_model/model.pnnx.bin', 'pnnxpy=best_ncnn_model/model_pnnx.py', 'pnnxonnx=best_ncnn_model/model.pnnx.onnx', 'fp16=0', 'device=cpu', 'inputshape="[1, 3, 640, 640]"']' died with <Signals.SIGABRT: 6>.
Does mnn provides the same kind of optimization on Android as compared to ncnn?
Does mnn provides the same kind of optimization on Android as compared to ncnn?
https://github.com/alibaba/MNN/blob/master/README.md
Lightweight Optimized for devices, no dependencies, can be easily deployed to mobile devices and a variety of embedded devices. iOS platform: static library size will full option for armv7+arm64 platforms is about 12MB, size increase of linked executables is about 2M. Android platform: core so size is about 800KB (armv7a - c++_shared).