refactor: uniformize options API

Introduces InferenceOptions for inference scripts.

The options can also be used by the server and be parsed either as json
or command line.

Options are also better organized.

New tests for inference GAN + diffusion

data/base_dataset.py

@@ -74,6 +74,10 @@ def modify_commandline_options(parser, is_train):
         """
         return parser
 
+    @staticmethod
+    def modify_commandline_options_train(parser):
+        return parser
+
     @abstractmethod
     def __len__(self):
         """Return the total number of images in the dataset."""

models/__init__.py

@@ -70,6 +70,12 @@ def get_option_setter(model_name):
     return model_class.modify_commandline_options
 
 
+def get_after_parse(model_name):
+    """Return the static method <modify_commandline_options> of the model class."""
+    model_class = find_model_using_name(model_name)
+    return model_class.after_parse
+
+
 def create_model(opt, rank):
     """Create a model given the option.
 

models/base_diffusion_model.py

@@ -14,6 +14,7 @@
 
 from util.diff_aug import DiffAugment
 from util.util import save_image, tensor2im
+from util.util import pairs_of_floats, pairs_of_ints, MAX_INT
 
 from .base_model import BaseModel
 from .modules.sam.sam_inference import (
@@ -35,6 +36,141 @@ class BaseDiffusionModel(BaseModel):
         -- <modify_commandline_options>:    (optionally) add model-specific options and set default options.
     """
 
+    @staticmethod
+    def modify_commandline_options(parser, is_train: bool = True):
+        print(parser)
+        if is_train:
+            BaseDiffusionModel.modify_commandline_options_train(parser)
+        else:
+            BaseDiffusionModel.modify_commandline_options_inference(parser)
+
+        return parser
+
+    @staticmethod
+    def modify_commandline_options_train(parser):
+        return parser
+
+    @staticmethod
+    def modify_commandline_options_inference(parser):
+        parser.add_argument(
+            "--crop_width",
+            default=-1,
+            type=int,
+            help="crop width added on each side of the bbox (optional)",
+        )
+        parser.add_argument(
+            "--crop_height",
+            default=-1,
+            type=int,
+            help="crop height added on each side of the bbox (optional)",
+        )
+
+        parser.add_argument(
+            "--bbox_width_factor",
+            type=float,
+            default=0.0,
+            help="bbox width added factor of original width",
+        )
+        parser.add_argument(
+            "--bbox_height_factor",
+            type=float,
+            default=0.0,
+            help="bbox height added factor of original height",
+        )
+
+        parser.add_argument(
+            "--sampling_steps", default=-1, type=int, help="number of sampling steps"
+        )
+        parser.add_argument(
+            "--seed", type=int, default=-1, help="random seed for reproducibility"
+        )
+
+        parser.add_argument(
+            "--mask_delta",
+            default=[[0]],
+            nargs="+",
+            type=pairs_of_ints,
+            help="mask offset to allow generation of a bigger object, format : width (x) height (y) for each class or only one size if square",
+        )
+        parser.add_argument(
+            "--mask_delta_ratio",
+            default=[[0]],
+            nargs="+",
+            type=pairs_of_floats,
+            help="ratio mask offset to allow generation of a bigger object, format : width (x),height (y) for each class or only one size if square",
+        )
+
+        parser.add_argument(
+            "--mask_square", action="store_true", help="whether to use square mask"
+        )
+
+        parser.add_argument("--name", help="generated img name", default="img")
+
+        parser.add_argument(
+            "--sampling_method",
+            type=str,
+            default="ddpm",
+            choices=["ddpm", "ddim"],
+            help="choose the sampling method between ddpm and ddim",
+        )
+
+        parser.add_argument(
+            "--cls_value",
+            type=int,
+            default=-1,
+            help="override input bbox classe for generation",
+        )
+
+        # last options
+        parser.add_argument("--previous_frame", help="image to transform", default=None)
+        parser.add_argument(
+            "--mask_in", help="mask used for image transformation", required=False
+        )
+        # XXX: put directly in the script?
+        parser.add_argument(
+            "--dir_out",
+            help="The directory where to output result images",
+            required=True,
+        )
+        parser.add_argument("--bbox_in", help="bbox file used for masking")
+
+        parser.add_argument(
+            "--nb_samples", help="nb of samples generated", type=int, default=1
+        )
+        parser.add_argument(
+            "--bbox_ref_id", help="bbox id to use", type=int, default=-1
+        )
+        parser.add_argument("--cond_in", help="conditionning image to use")
+        parser.add_argument("--cond_keep_ratio", action="store_true")
+        parser.add_argument("--cond_rotation", type=float, default=0)
+        parser.add_argument("--cond_persp_horizontal", type=float, default=0)
+        parser.add_argument("--cond_persp_vertical", type=float, default=0)
+
+        parser.add_argument(
+            "--alg_palette_guidance_scale",
+            type=float,
+            default=0.0,  # literature value: 0.2
+            help="scale for classifier-free guidance, default is conditional DDPM only",
+        )
+        parser.add_argument(
+            "--alg_palette_sketch_canny_thresholds",
+            type=int,
+            nargs="+",
+            default=[0, 255 * 3],
+            help="Canny thresholds",
+        )
+        parser.add_argument(
+            "--alg_palette_super_resolution_downsample",
+            action="store_true",
+            help="whether to downsample the image for super resolution",
+        )
+        parser.add_argument(
+            "--min_crop_bbox_ratio",
+            type=float,
+            help="minimum crop/bbox ratio, allows to add context when bbox is larger than crop",
+        )
+        return parser
+
     def __init__(self, opt, rank):
         """Initialize the BaseModel class.
 

models/base_model.py

@@ -358,6 +358,10 @@ def modify_commandline_options(parser, is_train):
         """
         return parser
 
+    @staticmethod
+    def modify_commandline_options_train(parser):
+        return parser
+
     def set_input(self, data):
         """Unpack input data from the dataloader and perform necessary pre-processing steps.
         Parameters:

models/cut_model.py

@@ -139,6 +139,10 @@ def modify_commandline_options(parser, is_train=True):
 
         return parser
 
+    @staticmethod
+    def after_parse(opt):
+        return opt
+
     def __init__(self, opt, rank):
         super().__init__(opt, rank)
 

models/cycle_gan_model.py

@@ -59,6 +59,10 @@ def modify_commandline_options(parser, is_train=True):
             )
         return parser
 
+    @staticmethod
+    def after_parse(opt):
+        return opt
+
     def __init__(self, opt, rank):
 
         super().__init__(opt, rank)

models/modules/sam/sam_inference.py

@@ -7,6 +7,7 @@
 import numpy as np
 import scipy
 import torch
+
 from mobile_sam.modeling import (
     ImageEncoderViT,
     MaskDecoder,