LNCT Excellence main gate, Bhopal (service road below NH 146)

The safety pyramid at this crossing

Not to scale. Near-misses happen far more often than collisions, so they can be counted now, before anyone is hurt (the traffic-safety pyramid: Laureshyn, Svensson & Hydén 2010).

Numbers road engineers use

• PV² and hourly rates need at least 15 minutes of video; this one is 8.6. Counted: 80 tracked crossings; 51 people crossed on foot; 55 moving vehicles.
• 69.6% of the time was inside gaps of 6 s or more with no moving vehicle where people crossed: 24 gaps, 9 of them 12 s or more, longest 63.2 s (the first and last are cut short by the video). UK Traffic Signs Manual ch. 6 (13.5.2): 4-6 s is an acceptable gap at urban speeds; other groups, such as children, may need 10-12 s or longer.
• 7 of 55 moving vehicles passed faster than 25 km/h, the speed India's Motor Vehicles (Driving) Regulations 2017 (reg. 16(5)) set past schools and hospitals where signed.
• SAM cost for this pack: about $12.35 at Meta's list prices (about $6.42 of it image detection), so about $86.50 per hour of footage.

Gaps in traffic

The whole 8:34 of video, left to right. Red, full height: a moving vehicle where people crossed. Blue: a gap of 12 s or more; amber: 6-12 s (see the UK Traffic Signs Manual line above). Hatched: cut short by the start or end of the video. Dots: near-misses (filled: serious; rings: close passes); the first 5 link to their cards below. Longest stretch with no gap of 6 s: 23.9 s.

Where it happened

Seen from above, in metres on the road (grid every 5 m; dashed: the measured area). Faint blue: where each person crossed, a point every 0.5 s. Blue: the pedestrian, red: the vehicle, 2 s either side. Filled dots: serious near-misses; rings: close passes. Under it, the street photo seen from above: the road surface is to scale, standing objects lean, and the dots are measured positions.

Where people crossed: 8 people walked 2 m or more, mostly across the road; 3 of them within the busiest 2 m stretch of road.

IRC:103-2012 cl. 6.7.1: "Pedestrian must be given the shortest possible direct route to cross the street". Cl. 6.1.9.2 (iii), schools and colleges: "Provision of guard-rails near schools where children would other-wise run straight into the road is essential. If there is a pedestrian crossing or a school crossing nearby, the guard-rails must be extended up to it."

Worst moments

0.0 s, serious near-miss, under 1.5 s

The pedestrian and the motorcycle were on the same spot at once.

#1 · serious · motorcycle at 18.7 km/h, 0.0 s from an adult

	Impact speed if hit	Adult death risk
As driven	18.7 km/h	0.5%
30 km/h limit	18.7 km/h	0.5%
20 km/h limit	18.7 km/h	0.5%

At 374.0 s in the video. Gap 0.42 m. The person and the motorcycle were on the spot at the same moment. Seen at the moment. The boxed figure is a person walking on foot with no bicycle or vehicle visible. It was measured for car fronts; for a motorcycle it is only a rough guide.

Heard (Muse Wonder): horn 1.0 s after.

0.9 s, serious near-miss, under 1.5 s

The motorcycle reached the pedestrian's spot 0.9 s after they left it.

#2 · serious · motorcycle at 18.3 km/h, 0.9 s from an adult

	Impact speed if hit	Adult death risk
As driven	18.3 km/h	0.5%
30 km/h limit	18.3 km/h	0.5%
20 km/h limit	18.3 km/h	0.5%

At 377.6 s in the video. Gap 0.73 m. The person was on the spot first; the motorcycle reached it 0.9 s later at 18.3 km/h. Seen at the moment. The boxed figure is a person walking on foot with no bicycle or vehicle visible. It was measured for car fronts; for a motorcycle it is only a rough guide.

1.9 s, close pass, 1.5-3.0 s

The motorcycle reached the pedestrian's spot 1.9 s after they left it.

#3 · close pass · motorcycle at 14.7 km/h, 1.9 s from an adult

	Impact speed if hit	Adult death risk
As driven	0.0 km/h	0.0%
30 km/h limit	0.0 km/h	0.0%
20 km/h limit	0.0 km/h	0.0%

At 150.0 s in the video. Gap 0.99 m. The person was on the spot first; the motorcycle reached it 1.9 s later at 14.7 km/h (13.0 km/h two seconds earlier: no slowdown measured). Seen at the moment. The boxed figure is walking on foot with both feet on the ground and not on any vehicle. It was measured for car fronts; for a motorcycle it is only a rough guide.

Heard (Muse Wonder): horn 0.8 s after.

2.1 s, close pass, 1.5-3.0 s

The motorcycle reached the pedestrian's spot 2.1 s after they left it.

#4 · close pass · motorcycle at 14.8 km/h, 2.1 s from an adult

	Impact speed if hit	Adult death risk
As driven	0.0 km/h	0.0%
30 km/h limit	0.0 km/h	0.0%
20 km/h limit	0.0 km/h	0.0%

At 376.6 s in the video. Gap 0.78 m. The person was on the spot first; the motorcycle reached it 2.1 s later at 14.8 km/h (9.1 km/h two seconds earlier: no slowdown measured). Seen at the moment. The boxed figure is a person walking on foot with no bicycle or vehicle visible. It was measured for car fronts; for a motorcycle it is only a rough guide.

Heard (Muse Wonder): horn 2.2 s before, horn 1.4 s after.

2.4 s, close pass, 1.5-3.0 s

The motorcycle reached the pedestrian's spot 2.4 s after they left it.

#5 · close pass · motorcycle at 10.2 km/h, 2.4 s from an adult

	Impact speed if hit	Adult death risk
As driven	0.0 km/h	0.0%
30 km/h limit	0.0 km/h	0.0%
20 km/h limit	0.0 km/h	0.0%

At 450.0 s in the video. Gap 0.95 m. The person was on the spot first; the motorcycle reached it 2.4 s later at 10.2 km/h. Seen at the moment. A person with a backpack is walking on foot across the road. It was measured for car fronts; for a motorcycle it is only a rough guide.

Heard (Muse Wonder): horn 2.0 s after.

Proposed fix: a painted zebra crossing at the gate with no hump and no speed breaker, and "College Zone 20 km/h" signs on both approaches

With 80 tracked crossings (51 people crossed on foot) and 55 moving vehicles in 8.6 minutes and 2 serious events and 3 close passes, a marked crossing with 20 km/h signs gives drivers and people a single expected place to cross.

AI-generated by Muse Image

Where the fix goes

Why the authority can act

“In school zones and other vulnerable corridors, authorities shall systematically identify high-risk stretches using accident data, and implement context-specific measures such as raised crossings, zebra markings, traffic calming features, and deployment of trained crossing guards.”Paragraph 35.5, page 33

“We, therefore, direct the road owning agencies in the 50 cities mentioned in Annexure 46 of the Report on Road Accidents, 2023 and the NHAI to draw up an action plan and start phase-wise survey to assess the requirement of additional pedestrian crossing facilities.”Paragraph 35.6, page 34

“State Governments/Municipal Authorities/NHAI are directed to create an online grievance redressal mechanism regarding footpaths which would include complaints of encroachments on footpaths, maintenance of footpaths and also suggestions for pedestrian crossing.”Paragraph 35.11, page 37

Bhopal is one of the 50 million-plus cities in Annexure 46 of MoRTH's Road Accidents in India 2023.

Paragraph 35.5 asks authorities to identify high-risk stretches using accident data. The near-misses measured here are not accidents: they are offered as further evidence about this crossing, alongside that data. Where the road authority has set up the online mechanism of paragraph 35.11, this pack can be filed there as a suggestion for a pedestrian crossing.

Source: Supreme Court of India, S. Rajaseekaran v. Union of India, Writ Petition (Civil) No. 295 of 2012, reportable order of 7 October 2025, 2025 INSC 1189 (PDF).

India's limit past a school or hospital where signs show it

“No driver shall drive at a speed exceeding twenty-five kilometre per hour or such lower speed as may be specified on the road signage while passing by a construction site or a school or a hospital, wherever indicated by signage, or on roads without footpaths and soft shoulders where pedestrians use a part of the carriageway to walk.”Regulation 16(5), PDF page 10

This rule applies where signs show it; we do not say any driver broke it.

Source: Motor Vehicles (Driving) Regulations 2017, Gazette of India Extraordinary No. 525, 23 June 2017, G.S.R. 634(E) (PDF).

Where this fits in a conflict study

This pack covers steps 5-7 (baseline counts, pre-intervention conflicts, analysis) of the CDC/FIA school-zone Traffic Conflict Technique toolkit, without the trained observers step 3 normally needs. Severity here is post-encroachment time and the measured speed, not the toolkit's observer severity scales, so the two do not compare one to one. Step 9 comes after the fix is in place.

With 3 of the 51 people who crossed having a moving vehicle through the same spot within 3 s (3 distinct vehicles), no after-film this long can show a drop (one-sided Fisher exact test at 5%, pre-registered as G7 in GATES.md): film 2 dismissals after the fix, from the same spot for as long each.

How this was measured

SAM 3.1 tracked the pedestrians and vehicles it found, at most 16 objects per video request (a note under the counts says when a request was full). Four points on the road turned pixels into metres. Muse Spark looked at the person in each near-miss, most dangerous first, and removed tracks that were not people; a card says when it did not look. A near-miss is a pedestrian and a moving vehicle on the same spot (within 1.0 m) less than 1.5 s apart (post-encroachment time); 1.5-3.0 s is reported as a close pass. Death risk: Rosén & Sander (2009), Accident Analysis & Prevention 41(3):536-542, for adults hit by a car front; children are not covered by that study. Speed-limit rows replay the same moment with the vehicle no faster than the limit, a 1.5 s reaction and 7.0 m/s² braking. Taped 8 m x 7 m rectangle (F1-F2 on the wall-side edge, N1-N2 on the gate-side edge), clicked on the stabilized frame at 300 s (F2 on the frame at 990 s, after a parked car left it). Camera 8.3 m above the road, hand-held; stabilize.py locks every frame to the 300 s frame.

Receipt: the analysis ran 1.1 minutes and took every SAM reply from cache, so it paid nothing new; Muse Spark looked at 32 people (32 of 32 verdicts reused from an earlier run: same crops, same replies); Muse Wonder listened around 5 near-misses; Muse Spark wrote the letter in 1 try; Muse Image drew 1 picture; the largest camera shift was 30 px (of 640), at 10:08. Dollar costs on this page are for SAM only.

Removed by Muse Spark: 28 tracks (track c482/0, not a person ("The cyan box contains only empty pavement with no person or rider v..."); track c396/0, rider ("The boxed person is seated driving a motor scooter with a passenger..."); track c389/0, rider ("The person is seated on and operating a motorcycle."); track c479/0, not a person ("The cyan box encloses only empty pavement with no person inside."); track c609/0, rider ("The box contains two people sitting on and riding a motorcycle."); track c113/0, rider ("The person is seated on a blue motor scooter as the driver."); track c141/0, rider ("The boxed person is seated as a pillion passenger on a moving motor..."); track c623/0, rider ("The boxed person is seated as a pillion passenger on a moving motor..."); track c124/0, rider ("The person is seated sideways on a motorcycle with other riders, no..."); track c100/0, rider ("The boxed person is seated on and steering a motorcycle."); track c512/0, rider ("The person in the box is sitting on and operating a motorcycle."); track c516/0, not a person ("The cyan box contains only empty pavement with no person present."); track c532/0, rider ("The person in the cyan box is sitting on and riding a red motorcycle."); track c117/0, rider ("The boxed persons are seated together on a motorcycle, not walking..."); track c128/0, not a person ("The box contains only empty road surface with no person present."); track c127/0, rider ("The person in the box is sitting as a passenger on a motorcycle."); track c541/0, rider ("The person is seated on a white motor scooter, not on foot."); track c123/0, rider ("The boxed person is driving a motorcycle with hands on handlebars."); track c508/0, rider ("The person is sitting on a motorcycle with a backpack."); track c629/0, rider ("The person in the box is seated on a motorcycle, not on foot."); track c591/0, rider ("The person is seated on a motorcycle, not walking on foot."); track c535/0, rider ("The boxed person is sitting as a pillion passenger on a red motorcy..."); track c122/0, rider ("Two people are seated together on a motor scooter, not walking on f..."); track c121/0, rider ("The box contains two people seated on a motorcycle, not on foot."); track c180/0, rider ("The person is sitting on and operating a motor scooter, not walking..."); track c418/0, rider ("The boxed person is sitting as a passenger on the motorcycle behind..."); track c715/0, not a person ("The cyan box contains only empty pavement with no visible person."); track c107/0, rider ("The boxed figure is a driver seated inside a van with arm through w...")). Ruled out by eye: track c116/0 (see GATES.md). Spark can be wrong.

The original video's SHA-256 is b30e9828eb13aa3aeb95199b21b6f1f2df409edc75b1d20fa6c2186961809900. Only minutes 1.6-10.2 of this 16.7-minute recording were analysed (seconds 95-609); times on this page count from second 95. SHA256SUMS in the shared folder lists every shared file's SHA-256: shasum -a 256 -c SHA256SUMS or python3 pack.py --verify FOLDER says if any changed. In India, Part A of the certificate for electronic evidence in the Schedule to the Bharatiya Sakshya Adhiniyam, 2023 (section 63(4)) asks for these hash values: the school fills and signs it.