Update modeling_internvl_chat.py

modeling_internvl_chat.py

@@ -5,13 +5,15 @@
 # --------------------------------------------------------
 import warnings
 from typing import Any, List, Optional, Tuple, Union
-
+from PIL import Image
 import torch.utils.checkpoint
 import transformers
 from torch import nn
 from torch.nn import CrossEntropyLoss
 from transformers import (AutoModel, GenerationConfig, LlamaForCausalLM,
                           LlamaTokenizer, Qwen2ForCausalLM)
+import torchvision.transforms as T
+from torchvision.transforms.functional import InterpolationMode
 from transformers.modeling_outputs import CausalLMOutputWithPast
 from transformers.modeling_utils import PreTrainedModel
 from transformers.utils import ModelOutput, logging
@@ -22,7 +24,8 @@ from .modeling_intern_vit import InternVisionModel, has_flash_attn
 from .modeling_internlm2 import InternLM2ForCausalLM
 
 logger = logging.get_logger(__name__)
-
+IMAGENET_MEAN = (0.485, 0.456, 0.406)
+IMAGENET_STD = (0.229, 0.224, 0.225)
 
 def version_cmp(v1, v2, op='eq'):
     import operator
@@ -31,6 +34,76 @@ def version_cmp(v1, v2, op='eq'):
     op_func = getattr(operator, op)
     return op_func(version.parse(v1), version.parse(v2))
 
+def build_transform(input_size):
+    MEAN, STD = IMAGENET_MEAN, IMAGENET_STD
+    transform = T.Compose([
+        T.Lambda(lambda img: img.convert('RGB') if img.mode != 'RGB' else img),
+        T.Resize((input_size, input_size), interpolation=InterpolationMode.BICUBIC),
+        T.ToTensor(),
+        T.Normalize(mean=MEAN, std=STD)
+    ])
+    return transform
+
+def find_closest_aspect_ratio(aspect_ratio, target_ratios, width, height, image_size):
+    best_ratio_diff = float('inf')
+    best_ratio = (1, 1)
+    area = width * height
+    for ratio in target_ratios:
+        target_aspect_ratio = ratio[0] / ratio[1]
+        ratio_diff = abs(aspect_ratio - target_aspect_ratio)
+        if ratio_diff < best_ratio_diff:
+            best_ratio_diff = ratio_diff
+            best_ratio = ratio
+        elif ratio_diff == best_ratio_diff:
+            if area > 0.5 * image_size * image_size * ratio[0] * ratio[1]:
+                best_ratio = ratio
+    return best_ratio
+
+def dynamic_preprocess(image, min_num=1, max_num=12, image_size=448, use_thumbnail=False):
+    orig_width, orig_height = image.size
+    aspect_ratio = orig_width / orig_height
+
+    # calculate the existing image aspect ratio
+    target_ratios = set(
+        (i, j) for n in range(min_num, max_num + 1) for i in range(1, n + 1) for j in range(1, n + 1) if
+        i * j <= max_num and i * j >= min_num)
+    target_ratios = sorted(target_ratios, key=lambda x: x[0] * x[1])
+
+    # find the closest aspect ratio to the target
+    target_aspect_ratio = find_closest_aspect_ratio(
+        aspect_ratio, target_ratios, orig_width, orig_height, image_size)
+
+    # calculate the target width and height
+    target_width = image_size * target_aspect_ratio[0]
+    target_height = image_size * target_aspect_ratio[1]
+    blocks = target_aspect_ratio[0] * target_aspect_ratio[1]
+
+    # resize the image
+    resized_img = image.resize((target_width, target_height))
+    processed_images = []
+    for i in range(blocks):
+        box = (
+            (i % (target_width // image_size)) * image_size,
+            (i // (target_width // image_size)) * image_size,
+            ((i % (target_width // image_size)) + 1) * image_size,
+            ((i // (target_width // image_size)) + 1) * image_size
+        )
+        # split the image
+        split_img = resized_img.crop(box)
+        processed_images.append(split_img)
+    assert len(processed_images) == blocks
+    if use_thumbnail and len(processed_images) != 1:
+        thumbnail_img = image.resize((image_size, image_size))
+        processed_images.append(thumbnail_img)
+    return processed_images
+
+def load_image(image_file, input_size=448, max_num=12):
+    image = Image.open(image_file).convert('RGB')
+    transform = build_transform(input_size=input_size)
+    images = dynamic_preprocess(image, image_size=input_size, use_thumbnail=True, max_num=max_num)
+    pixel_values = [transform(image) for image in images]
+    pixel_values = torch.stack(pixel_values)
+    return pixel_values
 
 class InternVLChatModel(PreTrainedModel):
     config_class = InternVLChatConfig
@@ -252,10 +325,25 @@ class InternVLChatModel(PreTrainedModel):
         responses = [response.split(template.sep)[0].strip() for response in responses]
         return responses
 
-    def chat(self, tokenizer, pixel_values, question, generation_config, history=None, return_history=False,
+    def chat(self, tokenizer, pixel_values, question, generation_config, history=None, return_history=False, image_dirs=None,
              num_patches_list=None, IMG_START_TOKEN='<img>', IMG_END_TOKEN='</img>', IMG_CONTEXT_TOKEN='<IMG_CONTEXT>',
              verbose=False):
-
+        if image_dirs is not None:
+            print("----------------------------------")
+            print("Using image_dirs to load images. 'pixel_values' and 'num_patches_list' will be ignored.")
+            print("You should provide all the previous image files and the current image file in the 'image_dirs' argument.")
+            print("----------------------------------")
+
+            if isinstance(image_dirs, str):
+                image_dirs = [image_dirs]
+            elif isinstance(image_dirs, list):
+                pass
+            else:
+                raise ValueError(f'Invalid image_dirs: {image_dirs}. It should be a string or a list of strings.')
+            image_values = [load_image(image_file, max_num=12).to(torch.float16).cuda() for image_file in image_dirs]
+            pixel_values = torch.cat(image_values, dim=0)
+            num_patches_list = [image_values[i].shape[0] for i in range(len(image_values))]
+            
         if history is None and pixel_values is not None and '<image>' not in question:
             question = '<image>\n' + question
 
@@ -360,4 +448,4 @@ class InternVLChatModel(PreTrainedModel):
         return self.language_model.get_input_embeddings()
 
     def get_output_embeddings(self):
-        return self.language_model.get_output_embeddings()
+        return self.language_model.get_output_embeddings()