Add safety module (#213)

* add SafetyChecker

* better name, fix checker

* add checker in main init

* remove from main init

* update logic to detect pipeline module

* style

* handle all safety logic in safety checker

* draw text

* can't draw

* small fixes

* treat special care as nsfw

* remove commented lines

* update safety checker

src/diffusers/pipeline_utils.py

@@ -42,6 +42,7 @@ LOADABLE_CLASSES = {
         "PreTrainedTokenizer": ["save_pretrained", "from_pretrained"],
         "PreTrainedTokenizerFast": ["save_pretrained", "from_pretrained"],
         "PreTrainedModel": ["save_pretrained", "from_pretrained"],
+        "FeatureExtractionMixin": ["save_pretrained", "from_pretrained"],
     },
 }
 
@@ -63,9 +64,9 @@ class DiffusionPipeline(ConfigMixin):
             library = module.__module__.split(".")[0]
 
             # check if the module is a pipeline module
-            pipeline_file = module.__module__.split(".")[-1]
             pipeline_dir = module.__module__.split(".")[-2]
-            is_pipeline_module = pipeline_file == "pipeline_" + pipeline_dir and hasattr(pipelines, pipeline_dir)
+            path = module.__module__.split(".")
+            is_pipeline_module = pipeline_dir in path and hasattr(pipelines, pipeline_dir)
 
             # if library is not in LOADABLE_CLASSES, then it is a custom module.
             # Or if it's a pipeline module, then the module is inside the pipeline

src/diffusers/pipelines/stable_diffusion/__init__.py

@@ -3,4 +3,4 @@ from ...utils import is_transformers_available
 
 
 if is_transformers_available():
-    from .pipeline_stable_diffusion import StableDiffusionPipeline
+    from .pipeline_stable_diffusion import StableDiffusionPipeline, StableDiffusionSafetyChecker

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py

@@ -4,11 +4,12 @@ from typing import List, Optional, Union
 import torch
 
 from tqdm.auto import tqdm
-from transformers import CLIPTextModel, CLIPTokenizer
+from transformers import CLIPFeatureExtractor, CLIPTextModel, CLIPTokenizer
 
 from ...models import AutoencoderKL, UNet2DConditionModel
 from ...pipeline_utils import DiffusionPipeline
 from ...schedulers import DDIMScheduler, LMSDiscreteScheduler, PNDMScheduler
+from .safety_checker import StableDiffusionSafetyChecker
 
 
 class StableDiffusionPipeline(DiffusionPipeline):
@@ -19,10 +20,20 @@ class StableDiffusionPipeline(DiffusionPipeline):
         tokenizer: CLIPTokenizer,
         unet: UNet2DConditionModel,
         scheduler: Union[DDIMScheduler, PNDMScheduler, LMSDiscreteScheduler],
+        safety_checker: StableDiffusionSafetyChecker,
+        feature_extractor: CLIPFeatureExtractor,
     ):
         super().__init__()
         scheduler = scheduler.set_format("pt")
-        self.register_modules(vae=vae, text_encoder=text_encoder, tokenizer=tokenizer, unet=unet, scheduler=scheduler)
+        self.register_modules(
+            vae=vae,
+            text_encoder=text_encoder,
+            tokenizer=tokenizer,
+            unet=unet,
+            scheduler=scheduler,
+            safety_checker=safety_checker,
+            feature_extractor=feature_extractor,
+        )
 
     @torch.no_grad()
     def __call__(
@@ -53,6 +64,7 @@ class StableDiffusionPipeline(DiffusionPipeline):
         self.unet.to(torch_device)
         self.vae.to(torch_device)
         self.text_encoder.to(torch_device)
+        self.safety_checker.to(torch_device)
 
         # get prompt text embeddings
         text_input = self.tokenizer(
@@ -136,7 +148,12 @@ class StableDiffusionPipeline(DiffusionPipeline):
 
         image = (image / 2 + 0.5).clamp(0, 1)
         image = image.cpu().permute(0, 2, 3, 1).numpy()
+
+        # run safety checker
+        safety_cheker_input = self.feature_extractor(self.numpy_to_pil(image), return_tensors="pt").to(torch_device)
+        image, has_nsfw_concept = self.safety_checker(images=image, clip_input=safety_cheker_input.pixel_values)
+
         if output_type == "pil":
             image = self.numpy_to_pil(image)
 
-        return {"sample": image}
+        return {"sample": image, "nsfw_content_detected": has_nsfw_concept}

src/diffusers/pipelines/stable_diffusion/safety_checker.py

@@ -0,0 +1,77 @@
+import numpy as np
+import torch
+import torch.nn as nn
+
+from transformers import CLIPConfig, CLIPVisionModel, PreTrainedModel
+
+from ...utils import logging
+
+
+logger = logging.get_logger(__name__)
+
+
+def cosine_distance(image_embeds, text_embeds):
+    normalized_image_embeds = nn.functional.normalize(image_embeds)
+    normalized_text_embeds = nn.functional.normalize(text_embeds)
+    return torch.mm(normalized_image_embeds, normalized_text_embeds.T)
+
+
+class StableDiffusionSafetyChecker(PreTrainedModel):
+    config_class = CLIPConfig
+
+    def __init__(self, config: CLIPConfig):
+        super().__init__(config)
+
+        self.vision_model = CLIPVisionModel(config.vision_config)
+        self.visual_projection = nn.Linear(config.vision_config.hidden_size, config.projection_dim, bias=False)
+
+        self.concept_embeds = nn.Parameter(torch.ones(17, config.projection_dim), requires_grad=False)
+        self.special_care_embeds = nn.Parameter(torch.ones(3, config.projection_dim), requires_grad=False)
+
+        self.register_buffer("concept_embeds_weights", torch.ones(17))
+        self.register_buffer("special_care_embeds_weights", torch.ones(3))
+
+    @torch.no_grad()
+    def forward(self, clip_input, images):
+        pooled_output = self.vision_model(clip_input)[1]  # pooled_output
+        image_embeds = self.visual_projection(pooled_output)
+
+        special_cos_dist = cosine_distance(image_embeds, self.special_care_embeds).cpu().numpy()
+        cos_dist = cosine_distance(image_embeds, self.concept_embeds).cpu().numpy()
+
+        result = []
+        batch_size = image_embeds.shape[0]
+        for i in range(batch_size):
+            result_img = {"special_scores": {}, "special_care": [], "concept_scores": {}, "bad_concepts": []}
+            adjustment = 0.05
+
+            for concet_idx in range(len(special_cos_dist[0])):
+                concept_cos = special_cos_dist[i][concet_idx]
+                concept_threshold = self.special_care_embeds_weights[concet_idx].item()
+                result_img["special_scores"][concet_idx] = round(concept_cos - concept_threshold + adjustment, 3)
+                if result_img["special_scores"][concet_idx] > 0:
+                    result_img["special_care"].append({concet_idx, result_img["special_scores"][concet_idx]})
+                    adjustment = 0.01
+
+            for concet_idx in range(len(cos_dist[0])):
+                concept_cos = cos_dist[i][concet_idx]
+                concept_threshold = self.concept_embeds_weights[concet_idx].item()
+                result_img["concept_scores"][concet_idx] = round(concept_cos - concept_threshold + adjustment, 3)
+                if result_img["concept_scores"][concet_idx] > 0:
+                    result_img["bad_concepts"].append(concet_idx)
+
+            result.append(result_img)
+
+        has_nsfw_concepts = [len(result[i]["bad_concepts"]) > 0 or i in range(len(result))]
+
+        for idx, has_nsfw_concept in enumerate(has_nsfw_concepts):
+            if has_nsfw_concept:
+                images[idx] = np.zeros(images[idx].shape)  # black image
+
+        if any(has_nsfw_concepts):
+            logger.warning(
+                "Potential NSFW content was detected in one or more images. A black image will be returned instead."
+                " Try again with a different prompt and/or seed."
+            )
+
+        return images, has_nsfw_concepts