Upload 29 files

stf/TEMP/.DS_Store

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d65165279105ca6773180500688df4bdc69a2c7b771752f0a46ef120b7fd8ec3
+size 6148

stf/TEMP/Cam2_2309071202_0012_Natural_Looped.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fed2d0f258d12caf91e35e0aeb5d7117fc21af282d2e6ed7f0437348079bece1
+size 231354130

stf/TEMP/front_config_v3.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:93b430287a06ce92cbb21a5cb96f576a0dea77cfd30c7b5ec3f8fbb6eea68b5e
+size 521

stf/convert.py

@@ -0,0 +1,20 @@
+import torch
+import numpy as np
+
+def convert():
+    state_dict = torch.load("mnist_cnn.pt")
+    
+    tensor = {
+        key: tensor.cpu().numpy() for key, tensor in state_dict.items()
+    }
+
+    for key, value in tensor.items():
+        print(key, value.shape)
+        
+    np.savez("mnist.npz", **tensor)
+    
+def main():
+    convert()
+
+if __name__ == "__main__":
+    main()

stf/front_config.json

@@ -0,0 +1,19 @@
+{
+  "name": "nasilhong_f_v1_front",
+  "batch_size": 16,
+  "num_workers": 16,
+  "mel_step_size": 108,
+  "mel_ps": 80,
+  "img_size": 352,
+  "mask_ver": "pwb_front_v39_1_2",
+  "num_ips": 2,
+  "mask_img_trsf_ver": 0,
+  "mel_trsf_ver": -1,
+  "mel_norm_ver": -1,
+  "lr": 1,
+  "crop_offset_y": 0.08,
+  "crop_margin": 0.35,
+  "crop_start_frame": 3,
+  "model_type": "stf_v3"
+}
+

stf/front_config_v3.json

@@ -0,0 +1,25 @@
+{
+  "name": "Ian_v3_front",
+  "batch_size": 16,
+  "num_workers": 16,
+  "mel_step_size": 108,
+  "mel_ps": 80,
+  "img_size": 352,
+  "mask_ver": "pwb_front_v39_1_wide",
+  "num_ips": 2,
+  "mask_img_trsf_ver": 0,
+  "mel_trsf_ver": -1,
+  "mel_norm_ver": -1,
+  "lr": 1,
+  "crop_offset_y": 0.08,
+  "crop_margin": 0.35,
+  "crop_start_frame": 3,
+  "model_type":"stf_v3",
+  "compose":"compose_smooth",
+  "image_resize":"True",
+  "compose_box": "face_only",
+  "compose_args": {
+      "blur_ratio": 0.3,
+     "dilate_ratio": 0.3
+  }
+}

stf/templates/front_one_piece_dress_nodded_cut.webm

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7f992d7c247e0b75995818d08b5e3a26ec117ccf4f47aac62bda126d444d29f2
+size 12800728

stf/test.py

@@ -0,0 +1,30 @@
+from time import time
+from datasets import load_dataset
+from faster_whisper import WhisperModel
+# from transformers import WhisperForConditionalGeneration, WhisperProcessor
+
+ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", cache_dir=".")
+
+# processor = WhisperProcessor.from_pretrained("openai/whisper-large-v3")
+# model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-large-v3").to("mps")
+model = WhisperModel("large-v3", device="cuda", compute_type="float16", download_root=".")
+
+audio_sample = ds[0]["audio"]
+waveform = audio_sample["array"]
+sampling_rate = audio_sample["sampling_rate"]
+
+tic = time()
+# input_features = processor(
+#     waveform, sampling_rate=sampling_rate, return_tensors="pt"
+# ).input_features
+segments, info = model.transcribe(waveform, beam_size=5)
+# predicted_ids = model.generate(input_features.to("mps"))
+
+# transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)
+
+toc = time()
+
+# print(transcription[0])
+for segment in segments:
+    print("[%.2fs -> %.2fs] %s" % (segment.start, segment.end, segment.text))
+print(toc - tic)

stf/works/preprocess/Ian_v3_front/.DS_Store

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b2bab5852e3a44786fd245355a3533ec9d21267ec58697abca3a1f9bb552775b
+size 6148

stf/works/preprocess/Ian_v3_front/Cam2_2309071202_0012_Natural_Looped.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:52b01f710ebe61088b0a5f22b7c4b2490023d370acc783573d935a4cbdadb8db
+size 511705088

stf/works/preprocess/Ian_v3_front/crop_video_Cam2_2309071202_0012_Natural_Looped.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:df3c585610ab66e232ae971be8ed43ce472b9a792fad05b7826c4396d2c64b77
+size 73588867

stf/works/preprocess/Ian_v3_front/df_anchor_i/Cam2_2309071202_0012_Natural_Looped_000.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:410850fc627f0c2cd6699f1825ccaee3e4a5ee26d54708e3bcceeeb511fd553e
+size 31559

stf/works/preprocess/Ian_v3_front/df_face_info/Cam2_2309071202_0012_Natural_Looped.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c73c0b3704e01331bcd7ee7a0e5fab0d55823d9158f8b81fa46fd3dad3241ea1
+size 117457

stf/works/preprocess/Ian_v3_front/metadata.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:615681285dec4805f5dd7886b7ee1c9a348ee0bb8efc38e91841bccf835bdccd
+size 42

stf/works/preprocess/nasilhong_f_v1_front/.DS_Store

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0c2786dddebbe68bf137be2a3005ceb4092317c279e75b605c5dc22ac16032e8
+size 6148

stf/works/preprocess/nasilhong_f_v1_front/crop_video_front_one_piece_dress_nodded_cut.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7bfefccf14d9b8411262eda85d2bac67679f7bd1015866a5fe8c92ae2a226a7d
+size 242568709

stf/works/preprocess/nasilhong_f_v1_front/df_anchor_i/front_golf_wear_nodded_cut_000.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9371a6b8c525a2eb9f90ca286145c97143e76f546afe24aa99e61ca33b34dcee
+size 38482

stf/works/preprocess/nasilhong_f_v1_front/df_anchor_i/front_one_piece_dress_nodded_cut_000.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dd4704c0a9c32b22b14deac1e966127d5a2727d957cfb1547f88b68ea18abdae
+size 38566

stf/works/preprocess/nasilhong_f_v1_front/df_face_info/front_golf_wear_nodded_cut.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:44408b86e82431f50b069d3a3d408e62706f106f0d7e79bcd7118723cda145fc
+size 143627

stf/works/preprocess/nasilhong_f_v1_front/df_face_info/front_one_piece_dress_nodded_cut.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05d953518ca36c9243168c56058b37a46aeb0cad940dec75cf0626ae75ec5f8f
+size 143591

stf/works/preprocess/nasilhong_f_v1_front/front_golf_wear_nodded_cut.webm

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bc3f6f0e5ab11b1db1c12fc397391e66b4152a0272b31528998b52fffc416fa9
+size 12855848

stf/works/preprocess/nasilhong_f_v1_front/front_one_piece_dress_nodded_cut.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb6e8fe920cdd7132c3824579159a77dd79f0b1afa56ad37ce3263fb018f8927
+size 902495052

stf/works/preprocess/nasilhong_f_v1_front/metadata.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:06c2997fe0bba05ca3922fd09d0ce435fb175d94de722d4027a6f82bdfeb6946
+size 42