happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

fix: test-1.jpg ORC issue with small (640) resolution 

setting min-resolution base (1024) max large (1280) for dynamic-resolution
This commit is contained in:
Saba Fallah 2025-12-10 20:20:55 +01:00
parent 016140699f
commit ed944cd25b
1 changed files with 85 additions and 160 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

83
tools/mtmd/clip.cpp
View File

@ -5213,7 +5213,7 @@ bool clip_image_preprocess(struct clip_ctx * ctx, const clip_image_u8 * img, str
case PROJECTOR_TYPE_DEEPSEEKOCR:
{
const int native_resolutions[] = {
512 /* tiny */, 640 /* small */, 1024 /* base */, 1280 /* large */
/* 512 tiny ,640 small ,*/ 1024 /* base */, 1280 /* large */
};
// original image size
const int orig_w = original_size.width;
@ -5228,7 +5228,7 @@ bool clip_image_preprocess(struct clip_ctx * ctx, const clip_image_u8 * img, str
int mode_i = 0;
int min_diff = orig_area;
for (int i = 0; i < 4; i++) {
for (int i = 0; i < 2; i++) {
int r = native_resolutions[i];
if (std::abs(orig_area - r * r) < min_diff) {
mode_i = i;
@ -5236,24 +5236,6 @@ bool clip_image_preprocess(struct clip_ctx * ctx, const clip_image_u8 * img, str
}
}
if (mode_i < 2) {
/* Native Resolution (Tiny/Small) */
const int image_size = native_resolutions[mode_i];
// Just resize the image to image_size × image_size
clip_image_u8_ptr resized_img(clip_image_u8_init());
img_tool::resize(*img, *resized_img,
clip_image_size{image_size, image_size},
img_tool::RESIZE_ALGO_BICUBIC_PILLOW, false, color); // Match PIL default
clip_image_f32_ptr res(clip_image_f32_init());
normalize_image_u8_to_f32(*resized_img, *res, params.image_mean, params.image_std);
res_imgs->entries.push_back(std::move(res));
res_imgs->grid_x = 1;
res_imgs->grid_y = 1;
}
else if (mode_i < 4) {
/* Native Resolution (Base/Large) */
const int image_size = native_resolutions[mode_i];
@ -5281,7 +5263,8 @@ bool clip_image_preprocess(struct clip_ctx * ctx, const clip_image_u8 * img, str
padded_img->buf.resize(image_size * image_size * 3); // black padding
// Fill with mean color
for (int i = 0; i < image_size * image_size; ++i) {
for (int i = 0; i < image_size * image_size; ++i)
{
padded_img->buf[i * 3 + 0] = pad_r;
padded_img->buf[i * 3 + 1] = pad_g;
padded_img->buf[i * 3 + 2] = pad_b;
@ -5310,66 +5293,8 @@ bool clip_image_preprocess(struct clip_ctx * ctx, const clip_image_u8 * img, str
res_imgs->grid_x = 1;
res_imgs->grid_y = 1;
}
else {
GGML_ABORT("DeepSeek-OCR hasn't supported Gundam/Gundam-Master yet");
/* Dynamic Resolution (Gundam/Gundam-Master) */
// configurable, or read from params
const int min_num = 2;
const int max_num = 9;
const int image_size = (mode_i == 4) ? 640 : 1024;
// original image size
const int orig_w = original_size.width;
const int orig_h = original_size.height;
// create overview image (thumbnail)
clip_image_u8_ptr overview_img(clip_image_u8_init());
img_tool::resize(*img, *overview_img, { image_size, image_size },
img_tool::RESIZE_ALGO_BICUBIC_PILLOW, true, color);
clip_image_f32_ptr overview_f32(clip_image_f32_init());
normalize_image_u8_to_f32(*overview_img, *overview_f32, params.image_mean, params.image_std);
res_imgs->entries.push_back(std::move(overview_f32));
// build candidate grids (cols, rows)
auto target_ratios = ds_build_target_ratios(min_num, max_num);
// pick the grid that best matches the original aspect ratio
const float aspect_ratio = static_cast<float>(orig_w) / static_cast<float>(orig_h);
auto best = ds_find_closest_ratio(aspect_ratio, target_ratios, orig_w, orig_h, image_size);
const int grid_cols = best.first; // how many tiles horizontally
const int grid_rows = best.second; // how many tiles vertically
// resize to refined size (no padding, direct resize)
clip_image_u8_ptr refined_img(clip_image_u8_init());
img_tool::resize(*img, *refined_img, { image_size * grid_cols, image_size * grid_rows },
img_tool::RESIZE_ALGO_BICUBIC_PILLOW, false);
// crop slices from the refined image
for (int r = 0; r < grid_rows; ++r) {
for (int c = 0; c < grid_cols; ++c) {
const int x = c * image_size;
const int y = r * image_size;
// crop the slice
clip_image_u8_ptr slice_img(clip_image_u8_init());
img_tool::crop(*refined_img, *slice_img, x, y, image_size, image_size);
// normalize and add to results
clip_image_f32_ptr slice_f32(clip_image_f32_init());
normalize_image_u8_to_f32(*slice_img, *slice_f32, params.image_mean, params.image_std);
res_imgs->entries.push_back(std::move(slice_f32));
}
}
// keep the grid info — the model may need to know how to reassemble / attend
res_imgs->grid_x = grid_cols;
res_imgs->grid_y = grid_rows;
}
}
break;
default:
LOG_ERR("%s: unsupported projector type %d\n", __func__, ctx->proj_type());
return false;