Upload model
Browse files- README.md +199 -0
- config.json +15 -0
- configuration_basnet.py +18 -0
- model.safetensors +3 -0
- modeling_basnet.py +481 -0
README.md
ADDED
|
@@ -0,0 +1,199 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
tags: []
|
| 4 |
+
---
|
| 5 |
+
|
| 6 |
+
# Model Card for Model ID
|
| 7 |
+
|
| 8 |
+
<!-- Provide a quick summary of what the model is/does. -->
|
| 9 |
+
|
| 10 |
+
|
| 11 |
+
|
| 12 |
+
## Model Details
|
| 13 |
+
|
| 14 |
+
### Model Description
|
| 15 |
+
|
| 16 |
+
<!-- Provide a longer summary of what this model is. -->
|
| 17 |
+
|
| 18 |
+
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
|
| 19 |
+
|
| 20 |
+
- **Developed by:** [More Information Needed]
|
| 21 |
+
- **Funded by [optional]:** [More Information Needed]
|
| 22 |
+
- **Shared by [optional]:** [More Information Needed]
|
| 23 |
+
- **Model type:** [More Information Needed]
|
| 24 |
+
- **Language(s) (NLP):** [More Information Needed]
|
| 25 |
+
- **License:** [More Information Needed]
|
| 26 |
+
- **Finetuned from model [optional]:** [More Information Needed]
|
| 27 |
+
|
| 28 |
+
### Model Sources [optional]
|
| 29 |
+
|
| 30 |
+
<!-- Provide the basic links for the model. -->
|
| 31 |
+
|
| 32 |
+
- **Repository:** [More Information Needed]
|
| 33 |
+
- **Paper [optional]:** [More Information Needed]
|
| 34 |
+
- **Demo [optional]:** [More Information Needed]
|
| 35 |
+
|
| 36 |
+
## Uses
|
| 37 |
+
|
| 38 |
+
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
|
| 39 |
+
|
| 40 |
+
### Direct Use
|
| 41 |
+
|
| 42 |
+
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
|
| 43 |
+
|
| 44 |
+
[More Information Needed]
|
| 45 |
+
|
| 46 |
+
### Downstream Use [optional]
|
| 47 |
+
|
| 48 |
+
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
|
| 49 |
+
|
| 50 |
+
[More Information Needed]
|
| 51 |
+
|
| 52 |
+
### Out-of-Scope Use
|
| 53 |
+
|
| 54 |
+
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
|
| 55 |
+
|
| 56 |
+
[More Information Needed]
|
| 57 |
+
|
| 58 |
+
## Bias, Risks, and Limitations
|
| 59 |
+
|
| 60 |
+
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
|
| 61 |
+
|
| 62 |
+
[More Information Needed]
|
| 63 |
+
|
| 64 |
+
### Recommendations
|
| 65 |
+
|
| 66 |
+
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
|
| 67 |
+
|
| 68 |
+
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
|
| 69 |
+
|
| 70 |
+
## How to Get Started with the Model
|
| 71 |
+
|
| 72 |
+
Use the code below to get started with the model.
|
| 73 |
+
|
| 74 |
+
[More Information Needed]
|
| 75 |
+
|
| 76 |
+
## Training Details
|
| 77 |
+
|
| 78 |
+
### Training Data
|
| 79 |
+
|
| 80 |
+
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
|
| 81 |
+
|
| 82 |
+
[More Information Needed]
|
| 83 |
+
|
| 84 |
+
### Training Procedure
|
| 85 |
+
|
| 86 |
+
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
|
| 87 |
+
|
| 88 |
+
#### Preprocessing [optional]
|
| 89 |
+
|
| 90 |
+
[More Information Needed]
|
| 91 |
+
|
| 92 |
+
|
| 93 |
+
#### Training Hyperparameters
|
| 94 |
+
|
| 95 |
+
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
|
| 96 |
+
|
| 97 |
+
#### Speeds, Sizes, Times [optional]
|
| 98 |
+
|
| 99 |
+
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
|
| 100 |
+
|
| 101 |
+
[More Information Needed]
|
| 102 |
+
|
| 103 |
+
## Evaluation
|
| 104 |
+
|
| 105 |
+
<!-- This section describes the evaluation protocols and provides the results. -->
|
| 106 |
+
|
| 107 |
+
### Testing Data, Factors & Metrics
|
| 108 |
+
|
| 109 |
+
#### Testing Data
|
| 110 |
+
|
| 111 |
+
<!-- This should link to a Dataset Card if possible. -->
|
| 112 |
+
|
| 113 |
+
[More Information Needed]
|
| 114 |
+
|
| 115 |
+
#### Factors
|
| 116 |
+
|
| 117 |
+
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
|
| 118 |
+
|
| 119 |
+
[More Information Needed]
|
| 120 |
+
|
| 121 |
+
#### Metrics
|
| 122 |
+
|
| 123 |
+
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
|
| 124 |
+
|
| 125 |
+
[More Information Needed]
|
| 126 |
+
|
| 127 |
+
### Results
|
| 128 |
+
|
| 129 |
+
[More Information Needed]
|
| 130 |
+
|
| 131 |
+
#### Summary
|
| 132 |
+
|
| 133 |
+
|
| 134 |
+
|
| 135 |
+
## Model Examination [optional]
|
| 136 |
+
|
| 137 |
+
<!-- Relevant interpretability work for the model goes here -->
|
| 138 |
+
|
| 139 |
+
[More Information Needed]
|
| 140 |
+
|
| 141 |
+
## Environmental Impact
|
| 142 |
+
|
| 143 |
+
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
|
| 144 |
+
|
| 145 |
+
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
|
| 146 |
+
|
| 147 |
+
- **Hardware Type:** [More Information Needed]
|
| 148 |
+
- **Hours used:** [More Information Needed]
|
| 149 |
+
- **Cloud Provider:** [More Information Needed]
|
| 150 |
+
- **Compute Region:** [More Information Needed]
|
| 151 |
+
- **Carbon Emitted:** [More Information Needed]
|
| 152 |
+
|
| 153 |
+
## Technical Specifications [optional]
|
| 154 |
+
|
| 155 |
+
### Model Architecture and Objective
|
| 156 |
+
|
| 157 |
+
[More Information Needed]
|
| 158 |
+
|
| 159 |
+
### Compute Infrastructure
|
| 160 |
+
|
| 161 |
+
[More Information Needed]
|
| 162 |
+
|
| 163 |
+
#### Hardware
|
| 164 |
+
|
| 165 |
+
[More Information Needed]
|
| 166 |
+
|
| 167 |
+
#### Software
|
| 168 |
+
|
| 169 |
+
[More Information Needed]
|
| 170 |
+
|
| 171 |
+
## Citation [optional]
|
| 172 |
+
|
| 173 |
+
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
|
| 174 |
+
|
| 175 |
+
**BibTeX:**
|
| 176 |
+
|
| 177 |
+
[More Information Needed]
|
| 178 |
+
|
| 179 |
+
**APA:**
|
| 180 |
+
|
| 181 |
+
[More Information Needed]
|
| 182 |
+
|
| 183 |
+
## Glossary [optional]
|
| 184 |
+
|
| 185 |
+
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
|
| 186 |
+
|
| 187 |
+
[More Information Needed]
|
| 188 |
+
|
| 189 |
+
## More Information [optional]
|
| 190 |
+
|
| 191 |
+
[More Information Needed]
|
| 192 |
+
|
| 193 |
+
## Model Card Authors [optional]
|
| 194 |
+
|
| 195 |
+
[More Information Needed]
|
| 196 |
+
|
| 197 |
+
## Model Card Contact
|
| 198 |
+
|
| 199 |
+
[More Information Needed]
|
config.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"BASNetModel"
|
| 4 |
+
],
|
| 5 |
+
"auto_map": {
|
| 6 |
+
"AutoConfig": "configuration_basnet.BASNetConfig",
|
| 7 |
+
"AutoModel": "modeling_basnet.BASNetModel"
|
| 8 |
+
},
|
| 9 |
+
"kernel_size": 3,
|
| 10 |
+
"model_type": "basnet",
|
| 11 |
+
"n_channels": 3,
|
| 12 |
+
"resnet_model": "microsoft/resnet-34",
|
| 13 |
+
"torch_dtype": "float32",
|
| 14 |
+
"transformers_version": "4.42.4"
|
| 15 |
+
}
|
configuration_basnet.py
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from transformers.configuration_utils import PretrainedConfig
|
| 2 |
+
|
| 3 |
+
|
| 4 |
+
class BASNetConfig(PretrainedConfig):
|
| 5 |
+
model_type = "basnet"
|
| 6 |
+
|
| 7 |
+
def __init__(
|
| 8 |
+
self,
|
| 9 |
+
resnet_model: str = "microsoft/resnet-34",
|
| 10 |
+
n_channels: int = 3,
|
| 11 |
+
kernel_size: int = 3,
|
| 12 |
+
**kwargs,
|
| 13 |
+
) -> None:
|
| 14 |
+
super().__init__(**kwargs)
|
| 15 |
+
self.resnet_model = resnet_model
|
| 16 |
+
self.n_channels = n_channels
|
| 17 |
+
|
| 18 |
+
self.kernel_size = 3
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:83db9a738691a9eca622ec38fac24b31e5b47121bec65570a3cf83f0f00ede32
|
| 3 |
+
size 348466168
|
modeling_basnet.py
ADDED
|
@@ -0,0 +1,481 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import logging
|
| 2 |
+
from typing import Optional, Tuple
|
| 3 |
+
|
| 4 |
+
import torch
|
| 5 |
+
import torch.nn as nn
|
| 6 |
+
import torchvision
|
| 7 |
+
from transformers.modeling_utils import PreTrainedModel
|
| 8 |
+
|
| 9 |
+
from .configuration_basnet import BASNetConfig
|
| 10 |
+
|
| 11 |
+
logger = logging.getLogger(__name__)
|
| 12 |
+
|
| 13 |
+
|
| 14 |
+
class RefUnet(nn.Module):
|
| 15 |
+
def __init__(self, in_ch: int, inc_ch: int) -> None:
|
| 16 |
+
super().__init__()
|
| 17 |
+
|
| 18 |
+
self.conv0 = nn.Conv2d(in_ch, inc_ch, kernel_size=3, padding=1)
|
| 19 |
+
|
| 20 |
+
self.conv1 = nn.Conv2d(inc_ch, 64, kernel_size=3, padding=1)
|
| 21 |
+
self.bn1 = nn.BatchNorm2d(64)
|
| 22 |
+
self.relu1 = nn.ReLU(inplace=True)
|
| 23 |
+
|
| 24 |
+
self.pool1 = nn.MaxPool2d(2, 2, ceil_mode=True)
|
| 25 |
+
|
| 26 |
+
self.conv2 = nn.Conv2d(64, 64, kernel_size=3, padding=1)
|
| 27 |
+
self.bn2 = nn.BatchNorm2d(64)
|
| 28 |
+
self.relu2 = nn.ReLU(inplace=True)
|
| 29 |
+
|
| 30 |
+
self.pool2 = nn.MaxPool2d(2, 2, ceil_mode=True)
|
| 31 |
+
|
| 32 |
+
self.conv3 = nn.Conv2d(64, 64, kernel_size=3, padding=1)
|
| 33 |
+
self.bn3 = nn.BatchNorm2d(64)
|
| 34 |
+
self.relu3 = nn.ReLU(inplace=True)
|
| 35 |
+
|
| 36 |
+
self.pool3 = nn.MaxPool2d(2, 2, ceil_mode=True)
|
| 37 |
+
|
| 38 |
+
self.conv4 = nn.Conv2d(64, 64, kernel_size=3, padding=1)
|
| 39 |
+
self.bn4 = nn.BatchNorm2d(64)
|
| 40 |
+
self.relu4 = nn.ReLU(inplace=True)
|
| 41 |
+
|
| 42 |
+
self.pool4 = nn.MaxPool2d(2, 2, ceil_mode=True)
|
| 43 |
+
|
| 44 |
+
#####
|
| 45 |
+
|
| 46 |
+
self.conv5 = nn.Conv2d(64, 64, kernel_size=3, padding=1)
|
| 47 |
+
self.bn5 = nn.BatchNorm2d(64)
|
| 48 |
+
self.relu5 = nn.ReLU(inplace=True)
|
| 49 |
+
|
| 50 |
+
#####
|
| 51 |
+
|
| 52 |
+
self.conv_d4 = nn.Conv2d(128, 64, kernel_size=3, padding=1)
|
| 53 |
+
self.bn_d4 = nn.BatchNorm2d(64)
|
| 54 |
+
self.relu_d4 = nn.ReLU(inplace=True)
|
| 55 |
+
|
| 56 |
+
self.conv_d3 = nn.Conv2d(128, 64, kernel_size=3, padding=1)
|
| 57 |
+
self.bn_d3 = nn.BatchNorm2d(64)
|
| 58 |
+
self.relu_d3 = nn.ReLU(inplace=True)
|
| 59 |
+
|
| 60 |
+
self.conv_d2 = nn.Conv2d(128, 64, kernel_size=3, padding=1)
|
| 61 |
+
self.bn_d2 = nn.BatchNorm2d(64)
|
| 62 |
+
self.relu_d2 = nn.ReLU(inplace=True)
|
| 63 |
+
|
| 64 |
+
self.conv_d1 = nn.Conv2d(128, 64, kernel_size=3, padding=1)
|
| 65 |
+
self.bn_d1 = nn.BatchNorm2d(64)
|
| 66 |
+
self.relu_d1 = nn.ReLU(inplace=True)
|
| 67 |
+
|
| 68 |
+
self.conv_d0 = nn.Conv2d(64, 1, kernel_size=3, padding=1)
|
| 69 |
+
|
| 70 |
+
self.upscore2 = nn.Upsample(
|
| 71 |
+
scale_factor=2, mode="bilinear", align_corners=False
|
| 72 |
+
)
|
| 73 |
+
# self.upscore2 = nn.Upsample(scale_factor=2, mode='bilinear')
|
| 74 |
+
|
| 75 |
+
def forward(self, x):
|
| 76 |
+
hx = x
|
| 77 |
+
hx = self.conv0(hx)
|
| 78 |
+
|
| 79 |
+
hx1 = self.relu1(self.bn1(self.conv1(hx)))
|
| 80 |
+
hx = self.pool1(hx1)
|
| 81 |
+
|
| 82 |
+
hx2 = self.relu2(self.bn2(self.conv2(hx)))
|
| 83 |
+
hx = self.pool2(hx2)
|
| 84 |
+
|
| 85 |
+
hx3 = self.relu3(self.bn3(self.conv3(hx)))
|
| 86 |
+
hx = self.pool3(hx3)
|
| 87 |
+
|
| 88 |
+
hx4 = self.relu4(self.bn4(self.conv4(hx)))
|
| 89 |
+
hx = self.pool4(hx4)
|
| 90 |
+
|
| 91 |
+
hx5 = self.relu5(self.bn5(self.conv5(hx)))
|
| 92 |
+
|
| 93 |
+
hx = self.upscore2(hx5)
|
| 94 |
+
|
| 95 |
+
d4 = self.relu_d4(self.bn_d4(self.conv_d4(torch.cat((hx, hx4), 1))))
|
| 96 |
+
hx = self.upscore2(d4)
|
| 97 |
+
|
| 98 |
+
d3 = self.relu_d3(self.bn_d3(self.conv_d3(torch.cat((hx, hx3), 1))))
|
| 99 |
+
hx = self.upscore2(d3)
|
| 100 |
+
|
| 101 |
+
d2 = self.relu_d2(self.bn_d2(self.conv_d2(torch.cat((hx, hx2), 1))))
|
| 102 |
+
hx = self.upscore2(d2)
|
| 103 |
+
|
| 104 |
+
d1 = self.relu_d1(self.bn_d1(self.conv_d1(torch.cat((hx, hx1), 1))))
|
| 105 |
+
|
| 106 |
+
residual = self.conv_d0(d1)
|
| 107 |
+
|
| 108 |
+
return x + residual
|
| 109 |
+
|
| 110 |
+
|
| 111 |
+
def conv3x3(in_planes, out_planes, stride=1) -> nn.Conv2d:
|
| 112 |
+
"3x3 convolution with padding"
|
| 113 |
+
return nn.Conv2d(
|
| 114 |
+
in_planes, out_planes, kernel_size=3, stride=stride, padding=1, bias=False
|
| 115 |
+
)
|
| 116 |
+
|
| 117 |
+
|
| 118 |
+
class BasicBlock(nn.Module):
|
| 119 |
+
expansion: int = 1
|
| 120 |
+
|
| 121 |
+
def __init__(self, inplanes: int, planes: int, stride: int = 1, downsample=None):
|
| 122 |
+
super(BasicBlock, self).__init__()
|
| 123 |
+
self.conv1 = conv3x3(inplanes, planes, stride)
|
| 124 |
+
self.bn1 = nn.BatchNorm2d(planes)
|
| 125 |
+
self.relu = nn.ReLU(inplace=True)
|
| 126 |
+
self.conv2 = conv3x3(planes, planes)
|
| 127 |
+
self.bn2 = nn.BatchNorm2d(planes)
|
| 128 |
+
self.downsample = downsample
|
| 129 |
+
self.stride = stride
|
| 130 |
+
|
| 131 |
+
def forward(self, x):
|
| 132 |
+
residual = x
|
| 133 |
+
|
| 134 |
+
out = self.conv1(x)
|
| 135 |
+
out = self.bn1(out)
|
| 136 |
+
out = self.relu(out)
|
| 137 |
+
|
| 138 |
+
out = self.conv2(out)
|
| 139 |
+
out = self.bn2(out)
|
| 140 |
+
|
| 141 |
+
if self.downsample is not None:
|
| 142 |
+
residual = self.downsample(x)
|
| 143 |
+
|
| 144 |
+
out += residual
|
| 145 |
+
out = self.relu(out)
|
| 146 |
+
|
| 147 |
+
return out
|
| 148 |
+
|
| 149 |
+
|
| 150 |
+
class BASNetModel(PreTrainedModel):
|
| 151 |
+
def __init__(self, config: BASNetConfig) -> None:
|
| 152 |
+
super().__init__(config)
|
| 153 |
+
|
| 154 |
+
resnet = torchvision.models.resnet34(
|
| 155 |
+
weights=torchvision.models.ResNet34_Weights.IMAGENET1K_V1
|
| 156 |
+
)
|
| 157 |
+
|
| 158 |
+
## -------------Encoder--------------
|
| 159 |
+
|
| 160 |
+
self.inconv = nn.Conv2d(
|
| 161 |
+
config.n_channels, 64, kernel_size=config.kernel_size, padding=1
|
| 162 |
+
)
|
| 163 |
+
self.inbn = nn.BatchNorm2d(64)
|
| 164 |
+
self.inrelu = nn.ReLU(inplace=True)
|
| 165 |
+
|
| 166 |
+
# stage 1
|
| 167 |
+
self.encoder1 = resnet.layer1 # 256
|
| 168 |
+
# stage 2
|
| 169 |
+
self.encoder2 = resnet.layer2 # 128
|
| 170 |
+
# stage 3
|
| 171 |
+
self.encoder3 = resnet.layer3 # 64
|
| 172 |
+
# stage 4
|
| 173 |
+
self.encoder4 = resnet.layer4 # 32
|
| 174 |
+
|
| 175 |
+
self.pool4 = nn.MaxPool2d(2, 2, ceil_mode=True)
|
| 176 |
+
|
| 177 |
+
# stage 5
|
| 178 |
+
self.resb5_1 = BasicBlock(512, 512)
|
| 179 |
+
self.resb5_2 = BasicBlock(512, 512)
|
| 180 |
+
self.resb5_3 = BasicBlock(512, 512) # 16
|
| 181 |
+
|
| 182 |
+
self.pool5 = nn.MaxPool2d(2, 2, ceil_mode=True)
|
| 183 |
+
|
| 184 |
+
# stage 6
|
| 185 |
+
self.resb6_1 = BasicBlock(512, 512)
|
| 186 |
+
self.resb6_2 = BasicBlock(512, 512)
|
| 187 |
+
self.resb6_3 = BasicBlock(512, 512) # 8
|
| 188 |
+
|
| 189 |
+
## -------------Bridge--------------
|
| 190 |
+
|
| 191 |
+
# stage Bridge
|
| 192 |
+
self.convbg_1 = nn.Conv2d(
|
| 193 |
+
512, 512, kernel_size=config.kernel_size, dilation=2, padding=2
|
| 194 |
+
) # 8
|
| 195 |
+
self.bnbg_1 = nn.BatchNorm2d(512)
|
| 196 |
+
self.relubg_1 = nn.ReLU(inplace=True)
|
| 197 |
+
self.convbg_m = nn.Conv2d(
|
| 198 |
+
512, 512, kernel_size=config.kernel_size, dilation=2, padding=2
|
| 199 |
+
)
|
| 200 |
+
self.bnbg_m = nn.BatchNorm2d(512)
|
| 201 |
+
self.relubg_m = nn.ReLU(inplace=True)
|
| 202 |
+
self.convbg_2 = nn.Conv2d(
|
| 203 |
+
512, 512, kernel_size=config.kernel_size, dilation=2, padding=2
|
| 204 |
+
)
|
| 205 |
+
self.bnbg_2 = nn.BatchNorm2d(512)
|
| 206 |
+
self.relubg_2 = nn.ReLU(inplace=True)
|
| 207 |
+
|
| 208 |
+
## -------------Decoder--------------
|
| 209 |
+
|
| 210 |
+
# stage 6d
|
| 211 |
+
self.conv6d_1 = nn.Conv2d(
|
| 212 |
+
1024, 512, kernel_size=config.kernel_size, padding=1
|
| 213 |
+
) # 16
|
| 214 |
+
self.bn6d_1 = nn.BatchNorm2d(512)
|
| 215 |
+
self.relu6d_1 = nn.ReLU(inplace=True)
|
| 216 |
+
|
| 217 |
+
self.conv6d_m = nn.Conv2d(
|
| 218 |
+
512, 512, kernel_size=config.kernel_size, dilation=2, padding=2
|
| 219 |
+
) ###
|
| 220 |
+
self.bn6d_m = nn.BatchNorm2d(512)
|
| 221 |
+
self.relu6d_m = nn.ReLU(inplace=True)
|
| 222 |
+
|
| 223 |
+
self.conv6d_2 = nn.Conv2d(
|
| 224 |
+
512, 512, kernel_size=config.kernel_size, dilation=2, padding=2
|
| 225 |
+
)
|
| 226 |
+
self.bn6d_2 = nn.BatchNorm2d(512)
|
| 227 |
+
self.relu6d_2 = nn.ReLU(inplace=True)
|
| 228 |
+
|
| 229 |
+
# stage 5d
|
| 230 |
+
self.conv5d_1 = nn.Conv2d(
|
| 231 |
+
1024, 512, kernel_size=config.kernel_size, padding=1
|
| 232 |
+
) # 16
|
| 233 |
+
self.bn5d_1 = nn.BatchNorm2d(512)
|
| 234 |
+
self.relu5d_1 = nn.ReLU(inplace=True)
|
| 235 |
+
|
| 236 |
+
self.conv5d_m = nn.Conv2d(
|
| 237 |
+
512, 512, kernel_size=config.kernel_size, padding=1
|
| 238 |
+
) ###
|
| 239 |
+
self.bn5d_m = nn.BatchNorm2d(512)
|
| 240 |
+
self.relu5d_m = nn.ReLU(inplace=True)
|
| 241 |
+
|
| 242 |
+
self.conv5d_2 = nn.Conv2d(512, 512, kernel_size=config.kernel_size, padding=1)
|
| 243 |
+
self.bn5d_2 = nn.BatchNorm2d(512)
|
| 244 |
+
self.relu5d_2 = nn.ReLU(inplace=True)
|
| 245 |
+
|
| 246 |
+
# stage 4d
|
| 247 |
+
self.conv4d_1 = nn.Conv2d(
|
| 248 |
+
1024, 512, kernel_size=config.kernel_size, padding=1
|
| 249 |
+
) # 32
|
| 250 |
+
self.bn4d_1 = nn.BatchNorm2d(512)
|
| 251 |
+
self.relu4d_1 = nn.ReLU(inplace=True)
|
| 252 |
+
|
| 253 |
+
self.conv4d_m = nn.Conv2d(
|
| 254 |
+
512, 512, kernel_size=config.kernel_size, padding=1
|
| 255 |
+
) ###
|
| 256 |
+
self.bn4d_m = nn.BatchNorm2d(512)
|
| 257 |
+
self.relu4d_m = nn.ReLU(inplace=True)
|
| 258 |
+
|
| 259 |
+
self.conv4d_2 = nn.Conv2d(512, 256, kernel_size=config.kernel_size, padding=1)
|
| 260 |
+
self.bn4d_2 = nn.BatchNorm2d(256)
|
| 261 |
+
self.relu4d_2 = nn.ReLU(inplace=True)
|
| 262 |
+
|
| 263 |
+
# stage 3d
|
| 264 |
+
self.conv3d_1 = nn.Conv2d(
|
| 265 |
+
512, 256, kernel_size=config.kernel_size, padding=1
|
| 266 |
+
) # 64
|
| 267 |
+
self.bn3d_1 = nn.BatchNorm2d(256)
|
| 268 |
+
self.relu3d_1 = nn.ReLU(inplace=True)
|
| 269 |
+
|
| 270 |
+
self.conv3d_m = nn.Conv2d(
|
| 271 |
+
256, 256, kernel_size=config.kernel_size, padding=1
|
| 272 |
+
) ###
|
| 273 |
+
self.bn3d_m = nn.BatchNorm2d(256)
|
| 274 |
+
self.relu3d_m = nn.ReLU(inplace=True)
|
| 275 |
+
|
| 276 |
+
self.conv3d_2 = nn.Conv2d(256, 128, kernel_size=config.kernel_size, padding=1)
|
| 277 |
+
self.bn3d_2 = nn.BatchNorm2d(128)
|
| 278 |
+
self.relu3d_2 = nn.ReLU(inplace=True)
|
| 279 |
+
|
| 280 |
+
# stage 2d
|
| 281 |
+
|
| 282 |
+
self.conv2d_1 = nn.Conv2d(
|
| 283 |
+
256, 128, kernel_size=config.kernel_size, padding=1
|
| 284 |
+
) # 128
|
| 285 |
+
self.bn2d_1 = nn.BatchNorm2d(128)
|
| 286 |
+
self.relu2d_1 = nn.ReLU(inplace=True)
|
| 287 |
+
|
| 288 |
+
self.conv2d_m = nn.Conv2d(
|
| 289 |
+
128, 128, kernel_size=config.kernel_size, padding=1
|
| 290 |
+
) ###
|
| 291 |
+
self.bn2d_m = nn.BatchNorm2d(128)
|
| 292 |
+
self.relu2d_m = nn.ReLU(inplace=True)
|
| 293 |
+
|
| 294 |
+
self.conv2d_2 = nn.Conv2d(128, 64, kernel_size=config.kernel_size, padding=1)
|
| 295 |
+
self.bn2d_2 = nn.BatchNorm2d(64)
|
| 296 |
+
self.relu2d_2 = nn.ReLU(inplace=True)
|
| 297 |
+
|
| 298 |
+
# stage 1d
|
| 299 |
+
self.conv1d_1 = nn.Conv2d(
|
| 300 |
+
128, 64, kernel_size=config.kernel_size, padding=1
|
| 301 |
+
) # 256
|
| 302 |
+
self.bn1d_1 = nn.BatchNorm2d(64)
|
| 303 |
+
self.relu1d_1 = nn.ReLU(inplace=True)
|
| 304 |
+
|
| 305 |
+
self.conv1d_m = nn.Conv2d(
|
| 306 |
+
64, 64, kernel_size=config.kernel_size, padding=1
|
| 307 |
+
) ###
|
| 308 |
+
self.bn1d_m = nn.BatchNorm2d(64)
|
| 309 |
+
self.relu1d_m = nn.ReLU(inplace=True)
|
| 310 |
+
|
| 311 |
+
self.conv1d_2 = nn.Conv2d(64, 64, kernel_size=config.kernel_size, padding=1)
|
| 312 |
+
self.bn1d_2 = nn.BatchNorm2d(64)
|
| 313 |
+
self.relu1d_2 = nn.ReLU(inplace=True)
|
| 314 |
+
|
| 315 |
+
## -------------Bilinear Upsampling--------------
|
| 316 |
+
self.upscore6 = nn.Upsample(
|
| 317 |
+
scale_factor=32, mode="bilinear", align_corners=False
|
| 318 |
+
) ###
|
| 319 |
+
self.upscore5 = nn.Upsample(
|
| 320 |
+
scale_factor=16, mode="bilinear", align_corners=False
|
| 321 |
+
)
|
| 322 |
+
self.upscore4 = nn.Upsample(
|
| 323 |
+
scale_factor=8, mode="bilinear", align_corners=False
|
| 324 |
+
)
|
| 325 |
+
self.upscore3 = nn.Upsample(
|
| 326 |
+
scale_factor=4, mode="bilinear", align_corners=False
|
| 327 |
+
)
|
| 328 |
+
self.upscore2 = nn.Upsample(
|
| 329 |
+
scale_factor=2, mode="bilinear", align_corners=False
|
| 330 |
+
)
|
| 331 |
+
|
| 332 |
+
# self.upscore6 = nn.Upsample(scale_factor=32, mode='bilinear') ###
|
| 333 |
+
# self.upscore5 = nn.Upsample(scale_factor=16, mode='bilinear')
|
| 334 |
+
# self.upscore4 = nn.Upsample(scale_factor=8, mode='bilinear')
|
| 335 |
+
# self.upscore3 = nn.Upsample(scale_factor=4, mode='bilinear')
|
| 336 |
+
# self.upscore2 = nn.Upsample(scale_factor=2, mode='bilinear')
|
| 337 |
+
|
| 338 |
+
## -------------Side Output--------------
|
| 339 |
+
self.outconvb = nn.Conv2d(512, 1, kernel_size=3, padding=1)
|
| 340 |
+
self.outconv6 = nn.Conv2d(512, 1, kernel_size=3, padding=1)
|
| 341 |
+
self.outconv5 = nn.Conv2d(512, 1, kernel_size=3, padding=1)
|
| 342 |
+
self.outconv4 = nn.Conv2d(256, 1, kernel_size=3, padding=1)
|
| 343 |
+
self.outconv3 = nn.Conv2d(128, 1, kernel_size=3, padding=1)
|
| 344 |
+
self.outconv2 = nn.Conv2d(64, 1, kernel_size=3, padding=1)
|
| 345 |
+
self.outconv1 = nn.Conv2d(64, 1, kernel_size=3, padding=1)
|
| 346 |
+
|
| 347 |
+
## -------------Refine Module-------------
|
| 348 |
+
self.refunet = RefUnet(1, 64)
|
| 349 |
+
|
| 350 |
+
self.post_init()
|
| 351 |
+
|
| 352 |
+
def forward(
|
| 353 |
+
self, pixel_values: torch.Tensor
|
| 354 |
+
) -> Tuple[
|
| 355 |
+
torch.Tensor,
|
| 356 |
+
torch.Tensor,
|
| 357 |
+
torch.Tensor,
|
| 358 |
+
torch.Tensor,
|
| 359 |
+
torch.Tensor,
|
| 360 |
+
torch.Tensor,
|
| 361 |
+
torch.Tensor,
|
| 362 |
+
torch.Tensor,
|
| 363 |
+
]:
|
| 364 |
+
hx = pixel_values
|
| 365 |
+
|
| 366 |
+
## -------------Encoder-------------
|
| 367 |
+
hx = self.inconv(hx)
|
| 368 |
+
hx = self.inbn(hx)
|
| 369 |
+
hx = self.inrelu(hx)
|
| 370 |
+
|
| 371 |
+
h1 = self.encoder1(hx) # 256
|
| 372 |
+
h2 = self.encoder2(h1) # 128
|
| 373 |
+
h3 = self.encoder3(h2) # 64
|
| 374 |
+
h4 = self.encoder4(h3) # 32
|
| 375 |
+
|
| 376 |
+
hx = self.pool4(h4) # 16
|
| 377 |
+
|
| 378 |
+
hx = self.resb5_1(hx)
|
| 379 |
+
hx = self.resb5_2(hx)
|
| 380 |
+
h5 = self.resb5_3(hx)
|
| 381 |
+
|
| 382 |
+
hx = self.pool5(h5) # 8
|
| 383 |
+
|
| 384 |
+
hx = self.resb6_1(hx)
|
| 385 |
+
hx = self.resb6_2(hx)
|
| 386 |
+
h6 = self.resb6_3(hx)
|
| 387 |
+
|
| 388 |
+
## -------------Bridge-------------
|
| 389 |
+
hx = self.relubg_1(self.bnbg_1(self.convbg_1(h6))) # 8
|
| 390 |
+
hx = self.relubg_m(self.bnbg_m(self.convbg_m(hx)))
|
| 391 |
+
hbg = self.relubg_2(self.bnbg_2(self.convbg_2(hx)))
|
| 392 |
+
|
| 393 |
+
## -------------Decoder-------------
|
| 394 |
+
|
| 395 |
+
hx = self.relu6d_1(self.bn6d_1(self.conv6d_1(torch.cat((hbg, h6), 1))))
|
| 396 |
+
hx = self.relu6d_m(self.bn6d_m(self.conv6d_m(hx)))
|
| 397 |
+
hd6 = self.relu6d_2(self.bn5d_2(self.conv6d_2(hx)))
|
| 398 |
+
|
| 399 |
+
hx = self.upscore2(hd6) # 8 -> 16
|
| 400 |
+
|
| 401 |
+
hx = self.relu5d_1(self.bn5d_1(self.conv5d_1(torch.cat((hx, h5), 1))))
|
| 402 |
+
hx = self.relu5d_m(self.bn5d_m(self.conv5d_m(hx)))
|
| 403 |
+
hd5 = self.relu5d_2(self.bn5d_2(self.conv5d_2(hx)))
|
| 404 |
+
|
| 405 |
+
hx = self.upscore2(hd5) # 16 -> 32
|
| 406 |
+
|
| 407 |
+
hx = self.relu4d_1(self.bn4d_1(self.conv4d_1(torch.cat((hx, h4), 1))))
|
| 408 |
+
hx = self.relu4d_m(self.bn4d_m(self.conv4d_m(hx)))
|
| 409 |
+
hd4 = self.relu4d_2(self.bn4d_2(self.conv4d_2(hx)))
|
| 410 |
+
|
| 411 |
+
hx = self.upscore2(hd4) # 32 -> 64
|
| 412 |
+
|
| 413 |
+
hx = self.relu3d_1(self.bn3d_1(self.conv3d_1(torch.cat((hx, h3), 1))))
|
| 414 |
+
hx = self.relu3d_m(self.bn3d_m(self.conv3d_m(hx)))
|
| 415 |
+
hd3 = self.relu3d_2(self.bn3d_2(self.conv3d_2(hx)))
|
| 416 |
+
|
| 417 |
+
hx = self.upscore2(hd3) # 64 -> 128
|
| 418 |
+
|
| 419 |
+
hx = self.relu2d_1(self.bn2d_1(self.conv2d_1(torch.cat((hx, h2), 1))))
|
| 420 |
+
hx = self.relu2d_m(self.bn2d_m(self.conv2d_m(hx)))
|
| 421 |
+
hd2 = self.relu2d_2(self.bn2d_2(self.conv2d_2(hx)))
|
| 422 |
+
|
| 423 |
+
hx = self.upscore2(hd2) # 128 -> 256
|
| 424 |
+
|
| 425 |
+
hx = self.relu1d_1(self.bn1d_1(self.conv1d_1(torch.cat((hx, h1), 1))))
|
| 426 |
+
hx = self.relu1d_m(self.bn1d_m(self.conv1d_m(hx)))
|
| 427 |
+
hd1 = self.relu1d_2(self.bn1d_2(self.conv1d_2(hx)))
|
| 428 |
+
|
| 429 |
+
## -------------Side Output-------------
|
| 430 |
+
db = self.outconvb(hbg)
|
| 431 |
+
db = self.upscore6(db) # 8->256
|
| 432 |
+
|
| 433 |
+
d6 = self.outconv6(hd6)
|
| 434 |
+
d6 = self.upscore6(d6) # 8->256
|
| 435 |
+
|
| 436 |
+
d5 = self.outconv5(hd5)
|
| 437 |
+
d5 = self.upscore5(d5) # 16->256
|
| 438 |
+
|
| 439 |
+
d4 = self.outconv4(hd4)
|
| 440 |
+
d4 = self.upscore4(d4) # 32->256
|
| 441 |
+
|
| 442 |
+
d3 = self.outconv3(hd3)
|
| 443 |
+
d3 = self.upscore3(d3) # 64->256
|
| 444 |
+
|
| 445 |
+
d2 = self.outconv2(hd2)
|
| 446 |
+
d2 = self.upscore2(d2) # 128->256
|
| 447 |
+
|
| 448 |
+
d1 = self.outconv1(hd1) # 256
|
| 449 |
+
|
| 450 |
+
## -------------Refine Module-------------
|
| 451 |
+
dout = self.refunet(d1) # 256
|
| 452 |
+
|
| 453 |
+
return (
|
| 454 |
+
torch.sigmoid(dout),
|
| 455 |
+
torch.sigmoid(d1),
|
| 456 |
+
torch.sigmoid(d2),
|
| 457 |
+
torch.sigmoid(d3),
|
| 458 |
+
torch.sigmoid(d4),
|
| 459 |
+
torch.sigmoid(d5),
|
| 460 |
+
torch.sigmoid(d6),
|
| 461 |
+
torch.sigmoid(db),
|
| 462 |
+
)
|
| 463 |
+
|
| 464 |
+
|
| 465 |
+
def convert_from_checkpoint(
|
| 466 |
+
repo_id: str, filename: str, config: Optional[BASNetConfig] = None
|
| 467 |
+
) -> BASNetModel:
|
| 468 |
+
from huggingface_hub import hf_hub_download
|
| 469 |
+
|
| 470 |
+
checkpoint_path = hf_hub_download(repo_id=repo_id, filename=filename)
|
| 471 |
+
|
| 472 |
+
config = config or BASNetConfig()
|
| 473 |
+
model = BASNetModel(config)
|
| 474 |
+
|
| 475 |
+
logger.info(f"Loading checkpoint from {checkpoint_path}")
|
| 476 |
+
state_dict = torch.load(checkpoint_path)
|
| 477 |
+
|
| 478 |
+
model.load_state_dict(state_dict, strict=True)
|
| 479 |
+
model.eval()
|
| 480 |
+
|
| 481 |
+
return model
|