Andrew Zisserman, Jaesung Huh, Dima Damen, Evangelos Kazakos, Jacob Chalk
We lifted 12 functions out of this paper's own repositories and ran 11 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.
| Repository | Role | Ran |
|---|---|---|
| JacobChalk/TIM | canonical | 11 of 12 |
| Function | Status | Where it lives |
|---|---|---|
| get_sinusoid_encoding_table | Ran | JacobChalk/TIM/feature_extractors/VideoMAE/modeling_finetune.py pointer only (licence: NONE) · get_code("da651e3979a18f84") |
| load_gt_segmentations | Ran | JacobChalk/TIM/detection/eval_detection/evaluate_detection_json.py pointer only (licence: NONE) · get_code("3777cf3d1476af2d") |
| load_gt_segmentations | Ran | JacobChalk/TIM/detection/eval_detection/evaluate_detection_json_ek100.py pointer only (licence: NONE) · get_code("5a45614aa6767966") |
| load_predicted_segmentations | Ran | JacobChalk/TIM/detection/eval_detection/evaluate_detection_json.py pointer only (licence: NONE) · get_code("91dbf25ea34df77b") |
| load_predicted_segmentations | Ran | JacobChalk/TIM/detection/eval_detection/evaluate_detection_json_ek100.py pointer only (licence: NONE) · get_code("67d66192df8ebf14") |
| sec_to_timestamp | Ran | JacobChalk/TIM/feature_extractors/make_framepickle.py pointer only (licence: NONE) · get_code("eb3884a04aa0a2ed") |
| sec_to_timestamp | Ran | JacobChalk/TIM/feature_extractors/make_videoinfo.py pointer only (licence: NONE) · get_code("2e4e22e309ff1d23") |
| timestamp_to_sec | Ran | JacobChalk/TIM/feature_extractors/make_framepickle.py pointer only (licence: NONE) · get_code("8a8e3d27500e4fd8") |
| timestamp_to_seconds | Ran | JacobChalk/TIM/feature_extractors/make_videoinfo.py pointer only (licence: NONE) · get_code("61dfbd1801e160f7") |
| timestamp_to_seconds | Ran | JacobChalk/TIM/detection/eval_detection/evaluate_detection_json.py pointer only (licence: NONE) · get_code("d45a7468dd11e23d") |
| train_class_batch | Ran | JacobChalk/TIM/feature_extractors/VideoMAE/engine_for_finetuning.py pointer only (licence: NONE) · get_code("93f25c853b7eeac5") |
| get_loss_scale_for_deepspeed | Not yet run | JacobChalk/TIM/feature_extractors/VideoMAE/engine_for_finetuning.py pointer only (licence: NONE) · get_code("29416043c7035e7c") |
Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.
Diverse actions give rise to rich audio-visual signals in long videos. Recent works showcase that the two modalities of audio and video exhibit different temporal extents of events and distinct labels. We address the interplay between the two modalities in long videos by explicitly modelling the temporal extents of audio and visual events. We propose the Time Interval Machine (TIM) where a modality-specific time interval poses as a query to a transformer encoder that ingests a long video input. The encoder then attends to the specified interval, as well as the surrounding context in both modalities, in order to recognise the ongoing action. We test TIM on three long audio-visual video datasets: EPIC-KITCHENS, Perception Test, and AVE, reporting state-of-the-art (SOTA) for recognition. On EPIC-KITCHENS, we beat previous SOTA that utilises LLMs and significantly larger pre-training by 2.9% top-1 action recognition accuracy. Additionally, we show that TIM can be adapted for action detection, using dense multi-scale interval queries, outperforming SOTA on EPIC-KITCHENS-100 for most metrics, and showing strong performance on the Perception Test. Our ablations show the critical role of integrating the two modalities and modelling their time intervals in achieving this performance. Code and models at: https://github.com/JacobChalk/TIM .
The same record, over MCP at https://syntology.ai/mcp:
get_harvested_code_for_paper("2404.05559")
get_code_for_paper("2404.05559")
have("2404.05559")
Connect an agent — have() is free.