diff --git a/python/sglang/srt/hardware_backend/npu/modules/qwen_vl_processor.py b/python/sglang/srt/hardware_backend/npu/modules/qwen_vl_processor.py index dfbbbf9273..c5c9ea5c24 100644 --- a/python/sglang/srt/hardware_backend/npu/modules/qwen_vl_processor.py +++ b/python/sglang/srt/hardware_backend/npu/modules/qwen_vl_processor.py @@ -8,10 +8,6 @@ from transformers.image_utils import ( SizeDict, get_image_size, ) -from transformers.models.qwen2_vl.image_processing_qwen2_vl import smart_resize -from transformers.models.qwen3_vl.video_processing_qwen3_vl import ( - smart_resize as smart_resize_video, -) from transformers.utils import TensorType from transformers.video_utils import group_videos_by_shape, reorder_videos @@ -86,19 +82,12 @@ def npu_wrapper_preprocess(func): ) resized_images_grouped = {} for shape, stacked_images in grouped_images.items(): - height, width = stacked_images.shape[-2:] if do_resize: - resized_height, resized_width = smart_resize( - height, - width, - factor=patch_size * merge_size, - min_pixels=size.shortest_edge, - max_pixels=size.longest_edge, - ) stacked_images = self.resize( - image=stacked_images, - size=SizeDict(height=resized_height, width=resized_width), + images=stacked_images, + size=size, resample=resample, + factor=patch_size * merge_size, ) resized_images_grouped[shape] = stacked_images resized_images = reorder_images(resized_images_grouped, grouped_images_index) @@ -193,26 +182,13 @@ def npu_wrapper_video_preprocess(func): resized_videos_grouped = {} for shape, stacked_videos in grouped_videos.items(): - B, T, C, H, W = stacked_videos.shape - num_frames, height, width = T, H, W if do_resize: - resized_height, resized_width = smart_resize_video( - num_frames=num_frames, - height=height, - width=width, - temporal_factor=temporal_patch_size, - factor=patch_size * merge_size, - min_pixels=size.shortest_edge, - max_pixels=size.longest_edge, - ) - stacked_videos = stacked_videos.view(B * T, C, H, W) stacked_videos = self.resize( - stacked_videos, - size=SizeDict(height=resized_height, width=resized_width), + videos=stacked_videos, + size=size, resample=resample, - ) - stacked_videos = stacked_videos.view( - B, T, C, resized_height, resized_width + factor=patch_size * merge_size, + temporal_factor=temporal_patch_size, ) resized_videos_grouped[shape] = stacked_videos resized_videos = reorder_videos(resized_videos_grouped, grouped_videos_index)