safety-checker (#3)

- safety_checker (aae8ffc757e1eb2325d09ab1b9dcf00fceaf44d5)


Co-authored-by: Radamés Ajna <radames@users.noreply.huggingface.co>

app.py

@@ -4,7 +4,10 @@ from diffusers import StableDiffusionXLPipeline, EulerDiscreteScheduler
 from huggingface_hub import hf_hub_download
 from safetensors.torch import load_file
 import spaces
+import os
+from PIL import Image
 
+SAFETY_CHECKER = os.environ.get("SAFETY_CHECKER", "0") == "1"
 
 # Constants
 base = "stabilityai/stable-diffusion-xl-base-1.0"
@@ -21,6 +24,27 @@ checkpoints = {
 if torch.cuda.is_available():
     pipe = StableDiffusionXLPipeline.from_pretrained(base, torch_dtype=torch.float16, variant="fp16").to("cuda")
 
+if SAFETY_CHECKER:
+    from safety_checker import StableDiffusionSafetyChecker
+    from transformers import CLIPFeatureExtractor
+
+    safety_checker = StableDiffusionSafetyChecker.from_pretrained(
+        "CompVis/stable-diffusion-safety-checker"
+    ).to("cuda")
+    feature_extractor = CLIPFeatureExtractor.from_pretrained(
+        "openai/clip-vit-base-patch32"
+    )
+
+    def check_nsfw_images(
+        images: list[Image.Image],
+    ) -> tuple[list[Image.Image], list[bool]]:
+        safety_checker_input = feature_extractor(images, return_tensors="pt").to("cuda")
+        has_nsfw_concepts = safety_checker(
+            images=[images],
+            clip_input=safety_checker_input.pixel_values.to("cuda")
+        )
+
+        return images, has_nsfw_concepts
 
 # Function 
 @spaces.GPU(enable_queue=True)
@@ -37,8 +61,16 @@ def generate_image(prompt, ckpt):
         pipe.scheduler = EulerDiscreteScheduler.from_config(pipe.scheduler.config, timestep_spacing="trailing")
         
     pipe.unet.load_state_dict(load_file(hf_hub_download(repo, checkpoint), device="cuda"))
-    image = pipe(prompt, num_inference_steps=num_inference_steps, guidance_scale=0).images[0]
-    return image
+    results = pipe(prompt, num_inference_steps=num_inference_steps, guidance_scale=0)
+
+    if SAFETY_CHECKER:
+        images, has_nsfw_concepts = check_nsfw_images(results.images)
+        if any(has_nsfw_concepts):
+            gr.Warning("NSFW content detected.")
+            return Image.new("RGB", (512, 512))
+        return images[0]
+    return results.images[0]
+
 
 
 # Gradio Interface

safety_checker.py

@@ -0,0 +1,137 @@
+# Copyright 2023 The HuggingFace Team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import numpy as np
+import torch
+import torch.nn as nn
+from transformers import CLIPConfig, CLIPVisionModel, PreTrainedModel
+
+
+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
+
+    _no_split_modules = ["CLIPEncoderLayer"]
+
+    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.concept_embeds_weights = nn.Parameter(torch.ones(17), requires_grad=False)
+        self.special_care_embeds_weights = nn.Parameter(
+            torch.ones(3), requires_grad=False
+        )
+
+    @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)
+
+        # we always cast to float32 as this does not cause significant overhead and is compatible with bfloat16
+        special_cos_dist = (
+            cosine_distance(image_embeds, self.special_care_embeds)
+            .cpu()
+            .float()
+            .numpy()
+        )
+        cos_dist = (
+            cosine_distance(image_embeds, self.concept_embeds).cpu().float().numpy()
+        )
+
+        result = []
+        batch_size = image_embeds.shape[0]
+        for i in range(batch_size):
+            result_img = {
+                "special_scores": {},
+                "special_care": [],
+                "concept_scores": {},
+                "bad_concepts": [],
+            }
+
+            # increase this value to create a stronger `nfsw` filter
+            # at the cost of increasing the possibility of filtering benign images
+            adjustment = 0.0
+
+            for concept_idx in range(len(special_cos_dist[0])):
+                concept_cos = special_cos_dist[i][concept_idx]
+                concept_threshold = self.special_care_embeds_weights[concept_idx].item()
+                result_img["special_scores"][concept_idx] = round(
+                    concept_cos - concept_threshold + adjustment, 3
+                )
+                if result_img["special_scores"][concept_idx] > 0:
+                    result_img["special_care"].append(
+                        {concept_idx, result_img["special_scores"][concept_idx]}
+                    )
+                    adjustment = 0.01
+
+            for concept_idx in range(len(cos_dist[0])):
+                concept_cos = cos_dist[i][concept_idx]
+                concept_threshold = self.concept_embeds_weights[concept_idx].item()
+                result_img["concept_scores"][concept_idx] = round(
+                    concept_cos - concept_threshold + adjustment, 3
+                )
+                if result_img["concept_scores"][concept_idx] > 0:
+                    result_img["bad_concepts"].append(concept_idx)
+
+            result.append(result_img)
+
+        has_nsfw_concepts = [len(res["bad_concepts"]) > 0 for res in result]
+
+        return has_nsfw_concepts
+
+    @torch.no_grad()
+    def forward_onnx(self, clip_input: torch.FloatTensor, images: torch.FloatTensor):
+        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)
+        cos_dist = cosine_distance(image_embeds, self.concept_embeds)
+
+        # increase this value to create a stronger `nsfw` filter
+        # at the cost of increasing the possibility of filtering benign images
+        adjustment = 0.0
+
+        special_scores = (
+            special_cos_dist - self.special_care_embeds_weights + adjustment
+        )
+        # special_scores = special_scores.round(decimals=3)
+        special_care = torch.any(special_scores > 0, dim=1)
+        special_adjustment = special_care * 0.01
+        special_adjustment = special_adjustment.unsqueeze(1).expand(
+            -1, cos_dist.shape[1]
+        )
+
+        concept_scores = (cos_dist - self.concept_embeds_weights) + special_adjustment
+        # concept_scores = concept_scores.round(decimals=3)
+        has_nsfw_concepts = torch.any(concept_scores > 0, dim=1)
+
+        images[has_nsfw_concepts] = 0.0  # black image
+
+        return images, has_nsfw_concepts