Commits · pesi/rtmo

Re-quantize models in FP16 in keep positional encoding in FP32 to avoid bad accuracy

f9a6075
verified

Luigi commited on Mar 27

Improve Debug Message Layout in Validate Funciton

e8f08e3

Luigi commited on Mar 26

Add appropriate pre- to input image and postprocessing to validate function

6dfe441

Luigi commited on Mar 26

Validate only on pose result, add visual check

b58a63b

Luigi commited on Mar 25

Add comparison between joint coordinated from original and converted models

ab55aa6

Luigi commited on Mar 25

Add 'rtol', 'atol' arguments to control tolerance in model validation

684ebde

Luigi commited on Mar 25

Fix Input Type Error raised in infer()

18eb8e3

Luigi commited on Mar 25

Add utility to convert ONNX model in FP32/16 mixed precision

9f60e86

Luigi commited on Mar 25

Load ./libmmdeploy_tensorrt_ops.so for TensorRT EP if available

1c5dc58

Luigi commited on Mar 19

Remove fix_batch_dimension_all.sh

01042ef

Luigi commited on Mar 19

Remove models with fixed batch size as dynamic batch is generally supported

ae1686f

Luigi commited on Mar 19

Apply constant folding

d474b05

Luigi commited on Mar 18

Reperform shape inference on all models

e7b68a5

Luigi commited on Mar 18

Bugfix: correct batch dimension fixation then regenerate ONNX models

e4e03fd

Luigi commited on Mar 16

Update README.md

bd45a81
verified

Luigi commited on Mar 9

Update README.md

0a36dd1
verified

Luigi commited on Mar 9

Add option --yolo_nas_pose, used to read YOLO NAS Pose model instead of RTMO

1f0f5d8

Luigi commited on Mar 8

Fix typo in model names mentioned in demo.sh demo_batch.sh

9133777

Luigi commited on Mar 8

Repair DEMO to run with TRT EP

c4f84a9

Luigi commited on Mar 8

Fix FP16 models having poor accuracy

25f1444

Luigi commited on Mar 8

Bugfix for batch demo: add padding if remaning number of frames is not equal to batch size

12e168d

Luigi commited on Mar 7

Make ONNX models compatible with ONNXruntime's TensorrtExecutionProvider

b75f05d

Luigi commited on Mar 7

Use RTMO_GPU_Batch to demo

3939a91

Luigi commited on Mar 5

Add new class 'RTMO_GPU_Batch' that can perform inference on batch of images

8072759

Luigi commited on Mar 5

Optimization for CUDAExecutionProvider

43a489d

Luigi commited on Mar 5

INT8 models obtained via dynamic quantization

e1ce29b
verified

Luigi commited on Mar 4

Update README.md

c4cff5e
verified

Luigi commited on Mar 1

Add a sample video

2e05515
verified

Luigi commited on Mar 1

RTMO ONNX Models from MMPOSE range from Tiny to Large

7ad7e4d
verified

Luigi commited on Mar 1

initial commit

ebf5707
verified

Luigi commited on Mar 1

pesi
/

rtmo

Commit History

Re-quantize models in FP16 in keep positional encoding in FP32 to avoid bad accuracy

f9a6075
verified

Improve Debug Message Layout in Validate Funciton

e8f08e3

Add appropriate pre- to input image and postprocessing to validate function

6dfe441

Validate only on pose result, add visual check

b58a63b

Add comparison between joint coordinated from original and converted models

ab55aa6

Add 'rtol', 'atol' arguments to control tolerance in model validation

684ebde

Fix Input Type Error raised in infer()

18eb8e3

Add utility to convert ONNX model in FP32/16 mixed precision

9f60e86

Load ./libmmdeploy_tensorrt_ops.so for TensorRT EP if available

1c5dc58

Remove fix_batch_dimension_all.sh

01042ef

Remove models with fixed batch size as dynamic batch is generally supported

ae1686f

Apply constant folding

d474b05

Reperform shape inference on all models

e7b68a5

Bugfix: correct batch dimension fixation then regenerate ONNX models

e4e03fd

Update README.md

bd45a81
verified

Update README.md

0a36dd1
verified

Add option --yolo_nas_pose, used to read YOLO NAS Pose model instead of RTMO

1f0f5d8

Fix typo in model names mentioned in demo.sh demo_batch.sh

9133777

Repair DEMO to run with TRT EP

c4f84a9

Fix FP16 models having poor accuracy

25f1444

Bugfix for batch demo: add padding if remaning number of frames is not equal to batch size

12e168d

Make ONNX models compatible with ONNXruntime's TensorrtExecutionProvider

b75f05d

Use RTMO_GPU_Batch to demo

3939a91

Add new class 'RTMO_GPU_Batch' that can perform inference on batch of images

8072759

Optimization for CUDAExecutionProvider

43a489d

INT8 models obtained via dynamic quantization

e1ce29b
verified

Update README.md

c4cff5e
verified

Add a sample video

2e05515
verified

RTMO ONNX Models from MMPOSE range from Tiny to Large

7ad7e4d
verified

initial commit

ebf5707
verified

Commit History

Re-quantize models in FP16 in keep positional encoding in FP32 to avoid bad accuracy f9a6075 verified

Improve Debug Message Layout in Validate Funciton e8f08e3

Add appropriate pre- to input image and postprocessing to validate function 6dfe441

Validate only on pose result, add visual check b58a63b

Add comparison between joint coordinated from original and converted models ab55aa6

Add 'rtol', 'atol' arguments to control tolerance in model validation 684ebde

Fix Input Type Error raised in infer() 18eb8e3

Add utility to convert ONNX model in FP32/16 mixed precision 9f60e86

Load ./libmmdeploy_tensorrt_ops.so for TensorRT EP if available 1c5dc58

Remove fix_batch_dimension_all.sh 01042ef

Remove models with fixed batch size as dynamic batch is generally supported ae1686f

Apply constant folding d474b05

Reperform shape inference on all models e7b68a5

Bugfix: correct batch dimension fixation then regenerate ONNX models e4e03fd

Update README.md bd45a81 verified

Update README.md 0a36dd1 verified

Add option --yolo_nas_pose, used to read YOLO NAS Pose model instead of RTMO 1f0f5d8

Fix typo in model names mentioned in demo.sh demo_batch.sh 9133777

Repair DEMO to run with TRT EP c4f84a9

Fix FP16 models having poor accuracy 25f1444

Bugfix for batch demo: add padding if remaning number of frames is not equal to batch size 12e168d

Make ONNX models compatible with ONNXruntime's TensorrtExecutionProvider b75f05d

Use RTMO_GPU_Batch to demo 3939a91

Add new class 'RTMO_GPU_Batch' that can perform inference on batch of images 8072759

Optimization for CUDAExecutionProvider 43a489d

INT8 models obtained via dynamic quantization e1ce29b verified

Update README.md c4cff5e verified

Add a sample video 2e05515 verified

RTMO ONNX Models from MMPOSE range from Tiny to Large 7ad7e4d verified

initial commit ebf5707 verified

Re-quantize models in FP16 in keep positional encoding in FP32 to avoid bad accuracy

f9a6075
verified

Improve Debug Message Layout in Validate Funciton

e8f08e3

Add appropriate pre- to input image and postprocessing to validate function

6dfe441

Validate only on pose result, add visual check

b58a63b

Add comparison between joint coordinated from original and converted models

ab55aa6

Add 'rtol', 'atol' arguments to control tolerance in model validation

684ebde

Fix Input Type Error raised in infer()

18eb8e3

Add utility to convert ONNX model in FP32/16 mixed precision

9f60e86

Load ./libmmdeploy_tensorrt_ops.so for TensorRT EP if available

1c5dc58

Remove fix_batch_dimension_all.sh

01042ef

Remove models with fixed batch size as dynamic batch is generally supported

ae1686f

Apply constant folding

d474b05

Reperform shape inference on all models

e7b68a5

Bugfix: correct batch dimension fixation then regenerate ONNX models

e4e03fd

Update README.md

bd45a81
verified

Update README.md

0a36dd1
verified

Add option --yolo_nas_pose, used to read YOLO NAS Pose model instead of RTMO

1f0f5d8

Fix typo in model names mentioned in demo.sh demo_batch.sh

9133777

Repair DEMO to run with TRT EP

c4f84a9

Fix FP16 models having poor accuracy

25f1444

Bugfix for batch demo: add padding if remaning number of frames is not equal to batch size

12e168d

Make ONNX models compatible with ONNXruntime's TensorrtExecutionProvider

b75f05d

Use RTMO_GPU_Batch to demo

3939a91

Add new class 'RTMO_GPU_Batch' that can perform inference on batch of images

8072759

Optimization for CUDAExecutionProvider

43a489d

INT8 models obtained via dynamic quantization

e1ce29b
verified

Update README.md

c4cff5e
verified

Add a sample video

2e05515
verified

RTMO ONNX Models from MMPOSE range from Tiny to Large

7ad7e4d
verified

initial commit

ebf5707
verified