export failure for tflite with options (--half and --int8)

[raccoondev85]

Search before asking

• I have searched the YOLOv5 issues and found no similar bug report.

YOLOv5 Component

Export

Bug

1. export failed when use gpu device for --half option

python export.py --half --weights ./export_tflite/yolov5n/best.pt --include tflite --device 0

LOG

export: data=data/coco128.yaml, weights=./export_tflite/yolov5n/best.pt, imgsz=[640, 640], batch_size=1, device=0, half=True, inplace=False, train=False, optimize=False, int8=False, dynamic=False, simplify=False, opset=13, topk_per_class=100, topk_all=100, iou_thres=0.45, conf_thres=0.25, include=['tflite']
YOLOv5 🚀 v6.0-42-g4c0982a torch 1.9.0+cu102 CUDA:0 (Quadro RTX 4000, 7982.3125MB)

Fusing layers...
Model Summary: 213 layers, 1764577 parameters, 0 gradients, 4.2 GFLOPs

PyTorch: starting from export_tflite/yolov5n/best.pt (14.7 MB)
2021-11-02 13:03:19.661502: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcudart.so.11.0

TensorFlow saved_model: starting export with tensorflow 2.4.1...

             from  n    params  module                                  arguments                     

TensorFlow saved_model: export failure: can't convert cuda:0 device type tensor to numpy. Use Tensor.cpu() to copy the tensor to host memory first.

TensorFlow Lite: starting export with tensorflow 2.4.1...

TensorFlow Lite: export failure: 'NoneType' object has no attribute 'call'

2. export failed when use --int8 option

python export.py --int8 --weights ./export_tflite/yolov5n/best.pt --include tflite

LOG

export: data=data/coco128.yaml, weights=./export_tflite/yolov5n/best.pt, imgsz=[640, 640], batch_size=1, device=cpu, half=False, inplace=False, train=False, optimize=False, int8=True, dynamic=False, simplify=False, opset=13, topk_per_class=100, topk_all=100, iou_thres=0.45, conf_thres=0.25, include=['tflite']
YOLOv5 🚀 v6.0-42-g4c0982a torch 1.9.0+cu102 CPU

Fusing layers...
Model Summary: 213 layers, 1764577 parameters, 0 gradients, 4.2 GFLOPs

PyTorch: starting from export_tflite/yolov5n/best.pt (14.7 MB)
2021-11-02 13:08:43.302145: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcudart.so.11.0

TensorFlow saved_model: starting export with tensorflow 2.4.1...

             from  n    params  module                                  arguments                     

2021-11-02 13:08:44.301976: I tensorflow/compiler/jit/xla_cpu_device.cc:41] Not creating XLA devices, tf_xla_enable_xla_devices not set
2021-11-02 13:08:44.303109: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcuda.so.1
2021-11-02 13:08:44.334274: E tensorflow/stream_executor/cuda/cuda_driver.cc:328] failed call to cuInit: CUDA_ERROR_NO_DEVICE: no CUDA-capable device is detected
2021-11-02 13:08:44.334350: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:169] retrieving CUDA diagnostic information for host: 19367d16ac94
2021-11-02 13:08:44.334358: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:176] hostname: 19367d16ac94
2021-11-02 13:08:44.335550: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:200] libcuda reported version is: 470.57.2
2021-11-02 13:08:44.335585: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:204] kernel reported version is: 470.57.2
2021-11-02 13:08:44.335591: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:310] kernel version seems to match DSO: 470.57.2
2021-11-02 13:08:44.336461: I tensorflow/core/platform/cpu_feature_guard.cc:142] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations: AVX2 AVX512F FMA
To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.
2021-11-02 13:08:44.338266: I tensorflow/compiler/jit/xla_gpu_device.cc:99] Not creating XLA devices, tf_xla_enable_xla_devices not set
0 -1 1 1760 models.common.Conv [3, 16, 6, 2, 2]
1 -1 1 4672 models.common.Conv [16, 32, 3, 2]
2 -1 1 4800 models.common.C3 [32, 32, 1]
3 -1 1 18560 models.common.Conv [32, 64, 3, 2]
4 -1 1 29184 models.common.C3 [64, 64, 2]
5 -1 1 73984 models.common.Conv [64, 128, 3, 2]
6 -1 1 156928 models.common.C3 [128, 128, 3]
7 -1 1 295424 models.common.Conv [128, 256, 3, 2]
8 -1 1 296448 models.common.C3 [256, 256, 1]
9 -1 1 164608 models.common.SPPF [256, 256, 5]
10 -1 1 33024 models.common.Conv [256, 128, 1, 1]
11 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
12 [-1, 6] 1 0 models.common.Concat [1]
13 -1 1 90880 models.common.C3 [256, 128, 1, False]
14 -1 1 8320 models.common.Conv [128, 64, 1, 1]
15 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
16 [-1, 4] 1 0 models.common.Concat [1]
17 -1 1 22912 models.common.C3 [128, 64, 1, False]
18 -1 1 36992 models.common.Conv [64, 64, 3, 2]
19 [-1, 14] 1 0 models.common.Concat [1]
20 -1 1 74496 models.common.C3 [128, 128, 1, False]
21 -1 1 147712 models.common.Conv [128, 128, 3, 2]
22 [-1, 10] 1 0 models.common.Concat [1]
23 -1 1 296448 models.common.C3 [256, 256, 1, False]
24 [17, 20, 23] 1 12177 models.yolo.Detect [4, [[10, 13, 16, 30, 33, 23], [30, 61, 62, 45, 59, 119], [116, 90, 156, 198, 373, 326]], [64, 128, 256], [640, 640]]
Model: "model"

---

Layer (type) Output Shape Param # Connected to

input_1 (InputLayer) [(1, 640, 640, 3)] 0

---

tf_conv (TFConv) (1, 320, 320, 16) 1744 input_1[0][0]

---

tf_conv_1 (TFConv) (1, 160, 160, 32) 4640 tf_conv[0][0]

---

tf_c3 (TFC3) (1, 160, 160, 32) 4704 tf_conv_1[0][0]

---

tf_conv_7 (TFConv) (1, 80, 80, 64) 18496 tf_c3[0][0]

---

tf_c3_1 (TFC3) (1, 80, 80, 64) 28928 tf_conv_7[0][0]

---

tf_conv_15 (TFConv) (1, 40, 40, 128) 73856 tf_c3_1[0][0]

---

tf_c3_2 (TFC3) (1, 40, 40, 128) 156288 tf_conv_15[0][0]

---

tf_conv_25 (TFConv) (1, 20, 20, 256) 295168 tf_c3_2[0][0]

---

tf_c3_3 (TFC3) (1, 20, 20, 256) 295680 tf_conv_25[0][0]

---

tfsppf (TFSPPF) (1, 20, 20, 256) 164224 tf_c3_3[0][0]

---

tf_conv_33 (TFConv) (1, 20, 20, 128) 32896 tfsppf[0][0]

---

tf_upsample (TFUpsample) (1, 40, 40, 128) 0 tf_conv_33[0][0]

---

tf_concat (TFConcat) (1, 40, 40, 256) 0 tf_upsample[0][0]
tf_c3_2[0][0]

---

tf_c3_4 (TFC3) (1, 40, 40, 128) 90496 tf_concat[0][0]

---

tf_conv_39 (TFConv) (1, 40, 40, 64) 8256 tf_c3_4[0][0]

---

tf_upsample_1 (TFUpsample) (1, 80, 80, 64) 0 tf_conv_39[0][0]

---

tf_concat_1 (TFConcat) (1, 80, 80, 128) 0 tf_upsample_1[0][0]
tf_c3_1[0][0]

---

tf_c3_5 (TFC3) (1, 80, 80, 64) 22720 tf_concat_1[0][0]

---

tf_conv_45 (TFConv) (1, 40, 40, 64) 36928 tf_c3_5[0][0]

---

tf_concat_2 (TFConcat) (1, 40, 40, 128) 0 tf_conv_45[0][0]
tf_conv_39[0][0]

---

tf_c3_6 (TFC3) (1, 40, 40, 128) 74112 tf_concat_2[0][0]

---

tf_conv_51 (TFConv) (1, 20, 20, 128) 147584 tf_c3_6[0][0]

---

tf_concat_3 (TFConcat) (1, 20, 20, 256) 0 tf_conv_51[0][0]
tf_conv_33[0][0]

---

tf_c3_7 (TFC3) (1, 20, 20, 256) 295680 tf_concat_3[0][0]

---

tf_detect (TFDetect) ((1, 25200, 9), [(1, 12177 tf_c3_5[0][0]
tf_c3_6[0][0]
tf_c3_7[0][0]

Total params: 1,764,577
Trainable params: 0
Non-trainable params: 1,764,577

---

2021-11-02 13:08:48.863933: W tensorflow/python/util/util.cc:348] Sets are not currently considered sequences, but this may change in the future, so consider avoiding using them.
Found untraced functions such as tf_conv_2_layer_call_and_return_conditional_losses, tf_conv_2_layer_call_fn, tf_conv_3_layer_call_and_return_conditional_losses, tf_conv_3_layer_call_fn, tf_conv_4_layer_call_and_return_conditional_losses while saving (showing 5 of 520). These functions will not be directly callable after loading.
Found untraced functions such as tf_conv_2_layer_call_and_return_conditional_losses, tf_conv_2_layer_call_fn, tf_conv_3_layer_call_and_return_conditional_losses, tf_conv_3_layer_call_fn, tf_conv_4_layer_call_and_return_conditional_losses while saving (showing 5 of 520). These functions will not be directly callable after loading.
Assets written to: export_tflite/yolov5n/best_saved_model/assets
TensorFlow saved_model: export success, saved as export_tflite/yolov5n/best_saved_model (62.2 MB)

TensorFlow Lite: starting export with tensorflow 2.4.1...
Found untraced functions such as tf_conv_2_layer_call_and_return_conditional_losses, tf_conv_2_layer_call_fn, tf_conv_3_layer_call_and_return_conditional_losses, tf_conv_3_layer_call_fn, tf_conv_4_layer_call_and_return_conditional_losses while saving (showing 5 of 520). These functions will not be directly callable after loading.
Found untraced functions such as tf_conv_2_layer_call_and_return_conditional_losses, tf_conv_2_layer_call_fn, tf_conv_3_layer_call_and_return_conditional_losses, tf_conv_3_layer_call_fn, tf_conv_4_layer_call_and_return_conditional_losses while saving (showing 5 of 520). These functions will not be directly callable after loading.
Assets written to: /tmp/tmpibgq2dv_/assets
2021-11-02 13:09:13.579602: I tensorflow/core/grappler/devices.cc:69] Number of eligible GPUs (core count >= 8, compute capability >= 0.0): 0
2021-11-02 13:09:13.579786: I tensorflow/core/grappler/clusters/single_machine.cc:356] Starting new session
2021-11-02 13:09:13.580027: I tensorflow/compiler/jit/xla_gpu_device.cc:99] Not creating XLA devices, tf_xla_enable_xla_devices not set
2021-11-02 13:09:13.605233: I tensorflow/core/platform/profile_utils/cpu_utils.cc:112] CPU Frequency: 3300000000 Hz
2021-11-02 13:09:13.622811: I tensorflow/core/grappler/optimizers/meta_optimizer.cc:928] Optimization results for grappler item: graph_to_optimize
function_optimizer: function_optimizer did nothing. time = 0.006ms.
function_optimizer: function_optimizer did nothing. time = 0ms.

2021-11-02 13:09:14.271508: W tensorflow/compiler/mlir/lite/python/tf_tfl_flatbuffer_helpers.cc:316] Ignored output_format.
2021-11-02 13:09:14.271571: W tensorflow/compiler/mlir/lite/python/tf_tfl_flatbuffer_helpers.cc:319] Ignored drop_control_dependency.
2021-11-02 13:09:14.311455: I tensorflow/compiler/jit/xla_gpu_device.cc:99] Not creating XLA devices, tf_xla_enable_xla_devices not set

TensorFlow Lite: export failure: too many values to unpack (expected 4)

Environment

1. YOLOv5 🚀 v6.0-42-g4c0982a torch 1.9.0+cu102 CUDA:0 (Quadro RTX 4000, 7982.3125MB)
2. YOLOv5 🚀 v6.0-42-g4c0982a torch 1.9.0+cu102 CPU
3. Ubuntu 18.04.5 LTS
4. python 3.8.8

Minimal Reproducible Example

No response

Additional

No response

Are you willing to submit a PR?

• Yes I'd like to help by submitting a PR!

[github-actions]

👋 Hello @raccoondev85, thank you for your interest in YOLOv5 🚀! Please visit our ⭐️ Tutorials to get started, where you can find quickstart guides for simple tasks like Custom Data Training all the way to advanced concepts like Hyperparameter Evolution.

If this is a 🐛 Bug Report, please provide screenshots and minimum viable code to reproduce your issue, otherwise we can not help you.

If this is a custom training ❓ Question, please provide as much information as possible, including dataset images, training logs, screenshots, and a public link to online W&B logging if available.

For business inquiries or professional support requests please visit https://ultralytics.com or email Glenn Jocher at glenn.jocher@ultralytics.com.

Requirements

Python>=3.6.0 with all requirements.txt installed including PyTorch>=1.7. To get started:

$ git clone https://github.com/ultralytics/yolov5
$ cd yolov5
$ pip install -r requirements.txt

Environments

YOLOv5 may be run in any of the following up-to-date verified environments (with all dependencies including CUDA/CUDNN, Python and PyTorch preinstalled):

• Google Colab and Kaggle notebooks with free GPU:
• Google Cloud Deep Learning VM. See GCP Quickstart Guide
• Amazon Deep Learning AMI. See AWS Quickstart Guide
• Docker Image. See Docker Quickstart Guide

Status

If this badge is green, all YOLOv5 GitHub Actions Continuous Integration (CI) tests are currently passing. CI tests verify correct operation of YOLOv5 training (train.py), validation (val.py), inference (detect.py) and export (export.py) on MacOS, Windows, and Ubuntu every 24 hours and on every commit.

[corehalt]

Works with tensorflow==2.6.0 but not with the today's latest version of tensorflow 2.6.1.
By default requirements.txt uses the latest available.

[glenn-jocher]

@corehalt am able to reproduce. Workaround is to use default settings. TFLite exports default to FP16, so the --half flag is not needed here. You can optionally apply --int8 to TFLite models. Also --device argument is not needed either.

[amanovv]

@glenn-jocher, tried on Colab, doesn't work same error

[glenn-jocher]

@amanovv thanks for the bug report! Yes I am able to reproduce this. Seems to be caused by PR #4854 from yesterday. Will push a fix soon.

[glenn-jocher]

@amanovv good news 😃! Your INT8 TFLite issue in #5446 (comment) may now be fixed ✅ in PR #5455. To receive this update:

• Git – git pull from within your yolov5/ directory or git clone https://github.com/ultralytics/yolov5 again
• PyTorch Hub – Force-reload model = torch.hub.load('ultralytics/yolov5', 'yolov5s', force_reload=True)
• Notebooks – View updated notebooks
• Docker – sudo docker pull ultralytics/yolov5:latest to update your image

Thank you for spotting this issue and informing us of the problem. Please let us know if this update resolves the issue for you, and feel free to inform us of any other issues you discover or feature requests that come to mind. Happy trainings with YOLOv5 🚀!