fix: pythorch 2.6 updates for all

detect.py

@@ -45,7 +45,7 @@ def detect(save_img=False):
     classify = False
     if classify:
         modelc = load_classifier(name='resnet101', n=2)  # initialize
-        modelc.load_state_dict(torch.load('weights/resnet101.pt', map_location=device)['model']).to(device).eval()
+        modelc.load_state_dict(torch.load('weights/resnet101.pt', map_location=device)['model'], weights_only=False).to(device).eval()
 
     # Set Dataloader
     vid_path, vid_writer = None, None

hubconf.py

@@ -37,7 +37,7 @@ def create(name, pretrained, channels, classes, autoshape):
         if pretrained:
             fname = f'{name}.pt'  # checkpoint filename
             attempt_download(fname)  # download if not found locally
-            ckpt = torch.load(fname, map_location=torch.device('cpu'))  # load
+            ckpt = torch.load(fname, map_location=torch.device('cpu'), weights_only=False)  # load
             msd = model.state_dict()  # model state_dict
             csd = ckpt['model'].float().state_dict()  # checkpoint state_dict as FP32
             csd = {k: v for k, v in csd.items() if msd[k].shape == v.shape}  # filter
@@ -65,7 +65,7 @@ def custom(path_or_model='path/to/model.pt', autoshape=True):
     Returns:
         pytorch model
     """
-    model = torch.load(path_or_model, map_location=torch.device('cpu')) if isinstance(path_or_model, str) else path_or_model  # load checkpoint
+    model = torch.load(path_or_model, map_location=torch.device('cpu'), weights_only=False) if isinstance(path_or_model, str) else path_or_model  # load checkpoint
     if isinstance(model, dict):
         model = model['ema' if model.get('ema') else 'model']  # load model
 

models/experimental.py

@@ -249,9 +249,9 @@ def attempt_load(weights, map_location=None):
     model = Ensemble()
     for w in weights if isinstance(weights, list) else [weights]:
         attempt_download(w)
-        ckpt = torch.load(w, map_location=map_location)  # load
+        ckpt = torch.load(w, map_location=map_location, weights_only=False)  # load
         model.append(ckpt['ema' if ckpt.get('ema') else 'model'].float().fuse().eval())  # FP32 model
-    
+
     # Compatibility updates
     for m in model.modules():
         if type(m) in [nn.Hardswish, nn.LeakyReLU, nn.ReLU, nn.ReLU6, nn.SiLU]:
@@ -260,7 +260,7 @@ def attempt_load(weights, map_location=None):
             m.recompute_scale_factor = None  # torch 1.11.0 compatibility
         elif type(m) is Conv:
             m._non_persistent_buffers_set = set()  # pytorch 1.6.0 compatibility
-    
+
     if len(model) == 1:
         return model[-1]  # return model
     else:

tools/reparameterization.ipynb

@@ -56,7 +56,7 @@
     "\n",
     "device = select_device('0', batch_size=1)\n",
     "# model trained by cfg/training/*.yaml\n",
-    "ckpt = torch.load('cfg/training/yolov7_training.pt', map_location=device)\n",
+    "ckpt = torch.load('cfg/training/yolov7_training.pt', map_location=device, weights_only=False)\n",
     "# reparameterized model in cfg/deploy/*.yaml\n",
     "model = Model('cfg/deploy/yolov7.yaml', ch=3, nc=80).to(device)\n",
     "\n",
@@ -118,7 +118,7 @@
     "\n",
     "device = select_device('0', batch_size=1)\n",
     "# model trained by cfg/training/*.yaml\n",
-    "ckpt = torch.load('cfg/training/yolov7x_trainig.pt', map_location=device)\n",
+    "ckpt = torch.load('cfg/training/yolov7x_trainig.pt', map_location=device, weights_only=False)\n",
     "# reparameterized model in cfg/deploy/*.yaml\n",
     "model = Model('cfg/deploy/yolov7x.yaml', ch=3, nc=80).to(device)\n",
     "\n",
@@ -180,7 +180,7 @@
     "\n",
     "device = select_device('0', batch_size=1)\n",
     "# model trained by cfg/training/*.yaml\n",
-    "ckpt = torch.load('cfg/training/yolov7-w6_trainig.pt', map_location=device)\n",
+    "ckpt = torch.load('cfg/training/yolov7-w6_trainig.pt', map_location=device, weights_only=False)\n",
     "# reparameterized model in cfg/deploy/*.yaml\n",
     "model = Model('cfg/deploy/yolov7-w6.yaml', ch=3, nc=80).to(device)\n",
     "\n",
@@ -266,7 +266,7 @@
     "\n",
     "device = select_device('0', batch_size=1)\n",
     "# model trained by cfg/training/*.yaml\n",
-    "ckpt = torch.load('cfg/training/yolov7-e6.pt', map_location=device)\n",
+    "ckpt = torch.load('cfg/training/yolov7-e6.pt', map_location=device, weights_only=False)\n",
     "# reparameterized model in cfg/deploy/*.yaml\n",
     "model = Model('cfg/deploy/yolov7-e6.yaml', ch=3, nc=80).to(device)\n",
     "\n",
@@ -352,7 +352,7 @@
     "\n",
     "device = select_device('0', batch_size=1)\n",
     "# model trained by cfg/training/*.yaml\n",
-    "ckpt = torch.load('cfg/training/yolov7-d6_trainig.pt', map_location=device)\n",
+    "ckpt = torch.load('cfg/training/yolov7-d6_trainig.pt', map_location=device, weights_only=False)\n",
     "# reparameterized model in cfg/deploy/*.yaml\n",
     "model = Model('cfg/deploy/yolov7-d6.yaml', ch=3, nc=80).to(device)\n",
     "\n",
@@ -438,7 +438,7 @@
     "\n",
     "device = select_device('0', batch_size=1)\n",
     "# model trained by cfg/training/*.yaml\n",
-    "ckpt = torch.load('cfg/training/yolov7-e6e_trainig.pt', map_location=device)\n",
+    "ckpt = torch.load('cfg/training/yolov7-e6e_trainig.pt', map_location=device, weights_only=False)\n",
     "# reparameterized model in cfg/deploy/*.yaml\n",
     "model = Model('cfg/deploy/yolov7-e6e.yaml', ch=3, nc=80).to(device)\n",
     "\n",

train_aux.py

@@ -68,7 +68,7 @@ def train(hyp, opt, device, tb_writer=None):
     loggers = {'wandb': None}  # loggers dict
     if rank in [-1, 0]:
         opt.hyp = hyp  # add hyperparameters
-        run_id = torch.load(weights).get('wandb_id') if weights.endswith('.pt') and os.path.isfile(weights) else None
+        run_id = torch.load(weights, weights_only=False).get('wandb_id') if weights.endswith('.pt') and os.path.isfile(weights) else None
         wandb_logger = WandbLogger(opt, Path(opt.save_dir).stem, run_id, data_dict)
         loggers['wandb'] = wandb_logger.wandb
         data_dict = wandb_logger.data_dict
@@ -84,7 +84,7 @@ def train(hyp, opt, device, tb_writer=None):
     if pretrained:
         with torch_distributed_zero_first(rank):
             attempt_download(weights)  # download if not found locally
-        ckpt = torch.load(weights, map_location=device)  # load checkpoint
+        ckpt = torch.load(weights, map_location=device, weights_only=False)  # load checkpoint
         model = Model(opt.cfg or ckpt['model'].yaml, ch=3, nc=nc, anchors=hyp.get('anchors')).to(device)  # create
         exclude = ['anchor'] if (opt.cfg or hyp.get('anchors')) and not opt.resume else []  # exclude keys
         state_dict = ckpt['model'].float().state_dict()  # to FP32
@@ -121,60 +121,60 @@ def train(hyp, opt, device, tb_writer=None):
         elif hasattr(v, 'weight') and isinstance(v.weight, nn.Parameter):
             pg1.append(v.weight)  # apply decay
         if hasattr(v, 'im'):
-            if hasattr(v.im, 'implicit'):           
+            if hasattr(v.im, 'implicit'):
                 pg0.append(v.im.implicit)
             else:
                 for iv in v.im:
                     pg0.append(iv.implicit)
         if hasattr(v, 'imc'):
-            if hasattr(v.imc, 'implicit'):           
+            if hasattr(v.imc, 'implicit'):
                 pg0.append(v.imc.implicit)
             else:
                 for iv in v.imc:
                     pg0.append(iv.implicit)
         if hasattr(v, 'imb'):
-            if hasattr(v.imb, 'implicit'):           
+            if hasattr(v.imb, 'implicit'):
                 pg0.append(v.imb.implicit)
             else:
                 for iv in v.imb:
                     pg0.append(iv.implicit)
         if hasattr(v, 'imo'):
-            if hasattr(v.imo, 'implicit'):           
+            if hasattr(v.imo, 'implicit'):
                 pg0.append(v.imo.implicit)
             else:
                 for iv in v.imo:
                     pg0.append(iv.implicit)
         if hasattr(v, 'ia'):
-            if hasattr(v.ia, 'implicit'):           
+            if hasattr(v.ia, 'implicit'):
                 pg0.append(v.ia.implicit)
             else:
                 for iv in v.ia:
                     pg0.append(iv.implicit)
         if hasattr(v, 'attn'):
-            if hasattr(v.attn, 'logit_scale'):   
+            if hasattr(v.attn, 'logit_scale'):
                 pg0.append(v.attn.logit_scale)
-            if hasattr(v.attn, 'q_bias'):   
+            if hasattr(v.attn, 'q_bias'):
                 pg0.append(v.attn.q_bias)
-            if hasattr(v.attn, 'v_bias'):  
+            if hasattr(v.attn, 'v_bias'):
                 pg0.append(v.attn.v_bias)
-            if hasattr(v.attn, 'relative_position_bias_table'):  
+            if hasattr(v.attn, 'relative_position_bias_table'):
                 pg0.append(v.attn.relative_position_bias_table)
         if hasattr(v, 'rbr_dense'):
-            if hasattr(v.rbr_dense, 'weight_rbr_origin'):  
+            if hasattr(v.rbr_dense, 'weight_rbr_origin'):
                 pg0.append(v.rbr_dense.weight_rbr_origin)
-            if hasattr(v.rbr_dense, 'weight_rbr_avg_conv'): 
+            if hasattr(v.rbr_dense, 'weight_rbr_avg_conv'):
                 pg0.append(v.rbr_dense.weight_rbr_avg_conv)
-            if hasattr(v.rbr_dense, 'weight_rbr_pfir_conv'):  
+            if hasattr(v.rbr_dense, 'weight_rbr_pfir_conv'):
                 pg0.append(v.rbr_dense.weight_rbr_pfir_conv)
-            if hasattr(v.rbr_dense, 'weight_rbr_1x1_kxk_idconv1'): 
+            if hasattr(v.rbr_dense, 'weight_rbr_1x1_kxk_idconv1'):
                 pg0.append(v.rbr_dense.weight_rbr_1x1_kxk_idconv1)
-            if hasattr(v.rbr_dense, 'weight_rbr_1x1_kxk_conv2'):   
+            if hasattr(v.rbr_dense, 'weight_rbr_1x1_kxk_conv2'):
                 pg0.append(v.rbr_dense.weight_rbr_1x1_kxk_conv2)
-            if hasattr(v.rbr_dense, 'weight_rbr_gconv_dw'):   
+            if hasattr(v.rbr_dense, 'weight_rbr_gconv_dw'):
                 pg0.append(v.rbr_dense.weight_rbr_gconv_dw)
-            if hasattr(v.rbr_dense, 'weight_rbr_gconv_pw'):   
+            if hasattr(v.rbr_dense, 'weight_rbr_gconv_pw'):
                 pg0.append(v.rbr_dense.weight_rbr_gconv_pw)
-            if hasattr(v.rbr_dense, 'vector'):   
+            if hasattr(v.rbr_dense, 'vector'):
                 pg0.append(v.rbr_dense.vector)
 
     if opt.adam:
@@ -642,12 +642,12 @@ if __name__ == '__main__':
                 'fliplr': (0, 0.0, 1.0),  # image flip left-right (probability)
                 'mosaic': (1, 0.0, 1.0),  # image mixup (probability)
                 'mixup': (1, 0.0, 1.0)}  # image mixup (probability)
-        
+
         with open(opt.hyp, errors='ignore') as f:
             hyp = yaml.safe_load(f)  # load hyps dict
             if 'anchors' not in hyp:  # anchors commented in hyp.yaml
                 hyp['anchors'] = 3
-                
+
         assert opt.local_rank == -1, 'DDP mode not implemented for --evolve'
         opt.notest, opt.nosave = True, True  # only test/save final epoch
         # ei = [isinstance(x, (int, float)) for x in hyp.values()]  # evolvable indices

utils/aws/resume.py

@@ -13,7 +13,7 @@ sys.path.append('./')  # to run '$ python *.py' files in subdirectories
 port = 0  # --master_port
 path = Path('').resolve()
 for last in path.rglob('*/**/last.pt'):
-    ckpt = torch.load(last)
+    ckpt = torch.load(last, weights_only=False)
     if ckpt['optimizer'] is None:
         continue
 

utils/datasets.py

@@ -361,7 +361,7 @@ class LoadImagesAndLabels(Dataset):  # for training/testing
         self.mosaic = self.augment and not self.rect  # load 4 images at a time into a mosaic (only during training)
         self.mosaic_border = [-img_size // 2, -img_size // 2]
         self.stride = stride
-        self.path = path        
+        self.path = path
         #self.albumentations = Albumentations() if augment else None
 
         try:
@@ -389,7 +389,7 @@ class LoadImagesAndLabels(Dataset):  # for training/testing
         self.label_files = img2label_paths(self.img_files)  # labels
         cache_path = (p if p.is_file() else Path(self.label_files[0]).parent).with_suffix('.cache')  # cached labels
         if cache_path.is_file():
-            cache, exists = torch.load(cache_path), True  # load
+            cache, exists = torch.load(cache_path, weights_only=False), True  # load
             #if cache['hash'] != get_hash(self.label_files + self.img_files) or 'version' not in cache:  # changed
             #    cache, exists = self.cache_labels(cache_path, prefix), False  # re-cache
         else:
@@ -576,8 +576,8 @@ class LoadImagesAndLabels(Dataset):  # for training/testing
                                                  scale=hyp['scale'],
                                                  shear=hyp['shear'],
                                                  perspective=hyp['perspective'])
-            
+
-            
+
             #img, labels = self.albumentations(img, labels)
 
             # Augment colorspace
@@ -586,9 +586,9 @@ class LoadImagesAndLabels(Dataset):  # for training/testing
             # Apply cutouts
             # if random.random() < 0.9:
             #     labels = cutout(img, labels)
-            
+
             if random.random() < hyp['paste_in']:
-                sample_labels, sample_images, sample_masks = [], [], [] 
+                sample_labels, sample_images, sample_masks = [], [], []
                 while len(sample_labels) < 30:
                     sample_labels_, sample_images_, sample_masks_ = load_samples(self, random.randint(0, len(self.labels) - 1))
                     sample_labels += sample_labels_
@@ -925,7 +925,7 @@ def remove_background(img, labels, segments):
         cv2.drawContours(im_new, [segments[j].astype(np.int32)], -1, (255, 255, 255), cv2.FILLED)
 
         result = cv2.bitwise_and(src1=img, src2=im_new)
-        
+
         i = result > 0  # pixels to replace
         img_new[i] = result[i]  # cv2.imwrite('debug.jpg', img)  # debug
 
@@ -942,19 +942,19 @@ def sample_segments(img, labels, segments, probability=0.5):
         h, w, c = img.shape  # height, width, channels
         for j in random.sample(range(n), k=round(probability * n)):
             l, s = labels[j], segments[j]
-            box = l[1].astype(int).clip(0,w-1), l[2].astype(int).clip(0,h-1), l[3].astype(int).clip(0,w-1), l[4].astype(int).clip(0,h-1) 
+            box = l[1].astype(int).clip(0,w-1), l[2].astype(int).clip(0,h-1), l[3].astype(int).clip(0,w-1), l[4].astype(int).clip(0,h-1)
-            
+
             #print(box)
             if (box[2] <= box[0]) or (box[3] <= box[1]):
                 continue
-            
+
             sample_labels.append(l[0])
-            
+
             mask = np.zeros(img.shape, np.uint8)
-            
+
             cv2.drawContours(mask, [segments[j].astype(np.int32)], -1, (255, 255, 255), cv2.FILLED)
             sample_masks.append(mask[box[1]:box[3],box[0]:box[2],:])
-            
+
             result = cv2.bitwise_and(src1=img, src2=mask)
             i = result > 0  # pixels to replace
             mask[i] = result[i]  # cv2.imwrite('debug.jpg', img)  # debug
@@ -1128,7 +1128,7 @@ def bbox_ioa(box1, box2):
 
     # Intersection over box2 area
     return inter_area / box2_area
-    
+
 
 def cutout(image, labels):
     # Applies image cutout augmentation https://arxiv.org/abs/1708.04552
@@ -1156,7 +1156,7 @@ def cutout(image, labels):
             labels = labels[ioa < 0.60]  # remove >60% obscured labels
 
     return labels
-    
+
 
 def pastein(image, labels, sample_labels, sample_images, sample_masks):
     # Applies image cutout augmentation https://arxiv.org/abs/1708.04552
@@ -1174,14 +1174,14 @@ def pastein(image, labels, sample_labels, sample_images, sample_masks):
         xmin = max(0, random.randint(0, w) - mask_w // 2)
         ymin = max(0, random.randint(0, h) - mask_h // 2)
         xmax = min(w, xmin + mask_w)
-        ymax = min(h, ymin + mask_h)   
+        ymax = min(h, ymin + mask_h)
-        
+
         box = np.array([xmin, ymin, xmax, ymax], dtype=np.float32)
         if len(labels):
-            ioa = bbox_ioa(box, labels[:, 1:5])  # intersection over area     
+            ioa = bbox_ioa(box, labels[:, 1:5])  # intersection over area
         else:
             ioa = np.zeros(1)
-        
+
         if (ioa < 0.30).all() and len(sample_labels) and (xmax > xmin+20) and (ymax > ymin+20):  # allow 30% obscuration of existing labels
             sel_ind = random.randint(0, len(sample_labels)-1)
             #print(len(sample_labels))
@@ -1194,7 +1194,7 @@ def pastein(image, labels, sample_labels, sample_images, sample_masks):
             r_scale = min((ymax-ymin)/hs, (xmax-xmin)/ws)
             r_w = int(ws*r_scale)
             r_h = int(hs*r_scale)
-            
+
             if (r_w > 10) and (r_h > 10):
                 r_mask = cv2.resize(sample_masks[sel_ind], (r_w, r_h))
                 r_image = cv2.resize(sample_images[sel_ind], (r_w, r_h))
@@ -1210,7 +1210,7 @@ def pastein(image, labels, sample_labels, sample_images, sample_masks):
                         labels = np.concatenate((labels, [[sample_labels[sel_ind], *box]]), 0)
                     else:
                         labels = np.array([[sample_labels[sel_ind], *box]])
-                              
+
                     image[ymin:ymin+r_h, xmin:xmin+r_w] = temp_crop
 
     return labels
@@ -1311,8 +1311,8 @@ def autosplit(path='../coco', weights=(0.9, 0.1, 0.0), annotated_only=False):
         if not annotated_only or Path(img2label_paths([str(img)])[0]).exists():  # check label
             with open(path / txt[i], 'a') as f:
                 f.write(str(img) + '\n')  # add image to txt file
-    
+
-    
+
 def load_segmentations(self, index):
     key = '/work/handsomejw66/coco17/' + self.img_files[index]
     #print(key)

utils/general.py

@@ -493,7 +493,7 @@ def box_giou(box1, box2):
 
     area1 = box_area(box1.T)
     area2 = box_area(box2.T)
-    
+
     inter = (torch.min(box1[:, None, 2:], box2[:, 2:]) - torch.max(box1[:, None, :2], box2[:, :2])).clamp(0).prod(2)
     union = (area1[:, None] + area2 - inter)
 
@@ -528,7 +528,7 @@ def box_ciou(box1, box2, eps: float = 1e-7):
 
     area1 = box_area(box1.T)
     area2 = box_area(box2.T)
-    
+
     inter = (torch.min(box1[:, None, 2:], box2[:, 2:]) - torch.max(box1[:, None, :2], box2[:, :2])).clamp(0).prod(2)
     union = (area1[:, None] + area2 - inter)
 
@@ -580,7 +580,7 @@ def box_diou(box1, box2, eps: float = 1e-7):
 
     area1 = box_area(box1.T)
     area2 = box_area(box2.T)
-    
+
     inter = (torch.min(box1[:, None, 2:], box2[:, 2:]) - torch.max(box1[:, None, :2], box2[:, :2])).clamp(0).prod(2)
     union = (area1[:, None] + area2 - inter)
 
@@ -799,7 +799,7 @@ def non_max_suppression_kpt(prediction, conf_thres=0.25, iou_thres=0.45, classes
 
 def strip_optimizer(f='best.pt', s=''):  # from utils.general import *; strip_optimizer()
     # Strip optimizer from 'f' to finalize training, optionally save as 's'
-    x = torch.load(f, map_location=torch.device('cpu'))
+    x = torch.load(f, map_location=torch.device('cpu'), weights_only=False)
     if x.get('ema'):
         x['model'] = x['ema']  # replace model with ema
     for k in 'optimizer', 'training_results', 'wandb_id', 'ema', 'updates':  # keys