quoccuongle/Grounded-SAM-2
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge pull request #12 from MorganTitcher/main

Browse Source 
fix: fixed typo in class name and in example notebooks
This commit is contained in:
Ren Tianhe
2024-08-14 14:47:17 +08:00
committed by GitHub
parent 9b8b41f04b f79b544fe6
commit 0df4f62ddc
3 changed files with 9 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
grounded_sam2_tracking_demo_with_continuous_id.py
View File

@@ -10,7 +10,7 @@ from transformers import AutoProcessor, AutoModelForZeroShotObjectDetection
from utils.track_utils import sample_points_from_masks from utils.track_utils import sample_points_from_masks
from utils.video_utils import create_video_from_images from utils.video_utils import create_video_from_images
from utils.common_utils import CommonUtils from utils.common_utils import CommonUtils
from utils.mask_dictionary_model import MaskDictionatyModel, ObjectInfo from utils.mask_dictionary_model import MaskDictionaryModel, ObjectInfo
import json import json
import copy import copy
@@ -70,7 +70,7 @@ frame_names.sort(key=lambda p: int(os.path.splitext(p)[0]))
inference_state = video_predictor.init_state(video_path=video_dir) inference_state = video_predictor.init_state(video_path=video_dir)
step = 10 # the step to sample frames for Grounding DINO predictor step = 10 # the step to sample frames for Grounding DINO predictor
sam2_masks = MaskDictionatyModel() sam2_masks = MaskDictionaryModel()
PROMPT_TYPE_FOR_VIDEO = "mask" # box, mask or point PROMPT_TYPE_FOR_VIDEO = "mask" # box, mask or point
objects_count = 0 objects_count = 0
@@ -85,7 +85,7 @@ for start_frame_idx in range(0, len(frame_names), step):
img_path = os.path.join(video_dir, frame_names[start_frame_idx]) img_path = os.path.join(video_dir, frame_names[start_frame_idx])
image = Image.open(img_path) image = Image.open(img_path)
image_base_name = frame_names[start_frame_idx].split(".")[0] image_base_name = frame_names[start_frame_idx].split(".")[0]
mask_dict = MaskDictionatyModel(promote_type = PROMPT_TYPE_FOR_VIDEO, mask_name = f"mask_{image_base_name}.npy") mask_dict = MaskDictionaryModel(promote_type = PROMPT_TYPE_FOR_VIDEO, mask_name = f"mask_{image_base_name}.npy")
# run Grounding DINO on the image # run Grounding DINO on the image
inputs = processor(images=image, text=text, return_tensors="pt").to(device) inputs = processor(images=image, text=text, return_tensors="pt").to(device)
@@ -155,7 +155,7 @@ for start_frame_idx in range(0, len(frame_names), step):
video_segments = {} # output the following {step} frames tracking masks video_segments = {} # output the following {step} frames tracking masks
for out_frame_idx, out_obj_ids, out_mask_logits in video_predictor.propagate_in_video(inference_state, max_frame_num_to_track=step, start_frame_idx=start_frame_idx): for out_frame_idx, out_obj_ids, out_mask_logits in video_predictor.propagate_in_video(inference_state, max_frame_num_to_track=step, start_frame_idx=start_frame_idx):
frame_masks = MaskDictionatyModel() frame_masks = MaskDictionaryModel()
for i, out_obj_id in enumerate(out_obj_ids): for i, out_obj_id in enumerate(out_obj_ids):
out_mask = (out_mask_logits[i] > 0.0) # .cpu().numpy() out_mask = (out_mask_logits[i] > 0.0) # .cpu().numpy()

8
grounded_sam2_tracking_demo_with_continuous_id_gd1.5.py
View File

@@ -16,7 +16,7 @@ from sam2.sam2_image_predictor import SAM2ImagePredictor
from transformers import AutoProcessor, AutoModelForZeroShotObjectDetection from transformers import AutoProcessor, AutoModelForZeroShotObjectDetection
from utils.video_utils import create_video_from_images from utils.video_utils import create_video_from_images
from utils.common_utils import CommonUtils from utils.common_utils import CommonUtils
from utils.mask_dictionary_model import MaskDictionatyModel, ObjectInfo from utils.mask_dictionary_model import MaskDictionaryModel, ObjectInfo
import json import json
import copy import copy
@@ -76,7 +76,7 @@ frame_names.sort(key=lambda p: int(os.path.splitext(p)[0]))
inference_state = video_predictor.init_state(video_path=video_dir) inference_state = video_predictor.init_state(video_path=video_dir)
step = 10 # the step to sample frames for Grounding DINO predictor step = 10 # the step to sample frames for Grounding DINO predictor
sam2_masks = MaskDictionatyModel() sam2_masks = MaskDictionaryModel()
PROMPT_TYPE_FOR_VIDEO = "mask" # box, mask or point PROMPT_TYPE_FOR_VIDEO = "mask" # box, mask or point
objects_count = 0 objects_count = 0
@@ -91,7 +91,7 @@ for start_frame_idx in range(0, len(frame_names), step):
img_path = os.path.join(video_dir, frame_names[start_frame_idx]) img_path = os.path.join(video_dir, frame_names[start_frame_idx])
image = Image.open(img_path) image = Image.open(img_path)
image_base_name = frame_names[start_frame_idx].split(".")[0] image_base_name = frame_names[start_frame_idx].split(".")[0]
mask_dict = MaskDictionatyModel(promote_type = PROMPT_TYPE_FOR_VIDEO, mask_name = f"mask_{image_base_name}.npy") mask_dict = MaskDictionaryModel(promote_type = PROMPT_TYPE_FOR_VIDEO, mask_name = f"mask_{image_base_name}.npy")
# run Grounding DINO 1.5 on the image # run Grounding DINO 1.5 on the image
@@ -174,7 +174,7 @@ for start_frame_idx in range(0, len(frame_names), step):
video_segments = {} # output the following {step} frames tracking masks video_segments = {} # output the following {step} frames tracking masks
for out_frame_idx, out_obj_ids, out_mask_logits in video_predictor.propagate_in_video(inference_state, max_frame_num_to_track=step, start_frame_idx=start_frame_idx): for out_frame_idx, out_obj_ids, out_mask_logits in video_predictor.propagate_in_video(inference_state, max_frame_num_to_track=step, start_frame_idx=start_frame_idx):
frame_masks = MaskDictionatyModel() frame_masks = MaskDictionaryModel()
for i, out_obj_id in enumerate(out_obj_ids): for i, out_obj_id in enumerate(out_obj_ids):
out_mask = (out_mask_logits[i] > 0.0) # .cpu().numpy() out_mask = (out_mask_logits[i] > 0.0) # .cpu().numpy()

2
utils/mask_dictionary_model.py
View File

@@ -7,7 +7,7 @@ import cv2
from dataclasses import dataclass, field from dataclasses import dataclass, field
@dataclass @dataclass
class MaskDictionatyModel: class MaskDictionaryModel:
mask_name:str = "" mask_name:str = ""
mask_height: int = 1080 mask_height: int = 1080
mask_width:int = 1920 mask_width:int = 1920